Skip to content
This repository has been archived by the owner on Aug 3, 2023. It is now read-only.

Latest commit

 

History

History
3550 lines (2162 loc) · 176 KB

CHANGELOG.md

File metadata and controls

3550 lines (2162 loc) · 176 KB
Raw

Changelog

1.19.11

  • Fixes

    • revert sites expiry stuff - caass, pull/2228

      After discussion with @threepointone and @petebacondarwin we've decided simply to revert back to the old behavior for workers-sites where we delete old assets, rather than marking them for expiration. See #2224 for our reasoning for this change.

1.19.10

v1.19.9

v1.19.8

  • Features

    • r2 bindings and subcommands - taylorlee, pull/2146

      Added r2 sub-commands to create/delete/list r2 bucket

v1.19.7

  • Fixes

    • Check for updates in the foreground, not the background - jyn514, pull/2166

      This avoids creating a new OpenSSL context while the main thread exits the process.

    • fix cli help typo - saikatdas0790, pull/2160

    • Improve heuristic for --openssl-legacy-provider (#2155) - eagleflo, pull/2172

      Improve the heuristic for detecting whether we need --openssl-legacy-provider added in #2116 by also checking whether the Node.js was actually built against OpenSSL 3 and has that option available.

    • Revert "Don't look for background updates unless Wrangler finished successfully" - jyn514, pull/2165

v1.19.6

  • Features

  • Fixes

    • Don't look for background updates unless Wrangler finished successfully - jyn514, pull/2150

      This works around a segfault due to OpenSSL's exit handlers not being thread-safe.

      See #1464 (comment) for an explanation and alternatives.

    • fix: incomplete binary with npm installation - 12f23eddde, pull/2149

      Closes #2148. This PR modifies binary-install.js (reference) to make sure the file stream is complete before the program finishes. I'm not a ... truncated

    • Get https websockets working - jyn514, pull/2153

      It turns out websocket upgrades with HTTP/2 require an HTTP extension, which Cloudflare doesn't currently support: https://datatracker.ietf.org/doc/html/rfc8441

      To avoid this, enable HTTP/1 for the remote client.

      This required an upd ... truncated

    • Get the audit CI job passing - jyn514, pull/2151

      Note that I didn't say "fix the vulnerabilities" - this just ignores the chrono and time vulnerabilities because they're both very hard to fix and not very common in practice.

      This uncovered a tokio vulnerability, which I've fixed by ... truncated

    • Proxy websocket connections when using authenticated (realish) preview - jyn514, pull/2135

      Previously, Wrangler would return a "101 Switching Protocols" response and then immediately close the TCP connection. This changes it to instead continue proxying the connection to the remote worker.

      This is simpler than `wrangler dev ... truncated

  • Maintenance

    • Update Rust toolchain to 1.57 - taylorlee, pull/2145

      I originally updated my toolchain to 1.56.1 on a local branch, since afaict rust-analyzer wasn't able to work with derived StructOpt proc-macros in 1.54. Since 1.57 was released today, I figured I'd update it all the way in this pr.

      foll ... truncated

v1.19.5

  • Features

    • add __STATIC_CONTENT_MANIFEST as a text module for modules workers - g45t345rt, pull/2126

      The __STATIC_CONTENT_MANIFEST is not accessible when using modules workers [sites]

      This pr makes it possible to import the content of manifest as text.

      It is also in favor of storing the manifest in vars https://github.com/cloudf ... truncated

  • Maintenance

    • Bump os-version to 0.2.0 - a1ien, pull/2122

      In 0.2 os-version use new stable version of plist In plist 1.2.2(which is not published yet) remove chrono dependency and use time 0.3 which fix part of #2117

    • Fix missing fmt argument in error message - dtolnay, pull/2118

      • The first commit fills in a missing argument to {} in config.rs. Without this, the error message would literally be 'Your token has status "{}"' with curly braces in it instead of the token status.

      • The second commit closes an incon ... truncated

    • Fix some migration errors - xortive, pull/2127

      • Don't include migrations in Target for preview/dev uploads

    • Fixes wrangler.toml being incorrectly inserted into existing directory - Scotsoo, pull/2110

      Fixes #2051 by passing new_name rather than name to config_path path

    • Remove serde-hjson transitive dependency - a1ien, pull/2123

      serde-hjson crate is not maintained anymore With removing this dependency we also remove duplicate of old serde Also in new version of config crate hjson also removed

    • Remove unneeded heartbeat logic with sketchy time addition - nataliescottdavidson, pull/2129

    • Remove unused eventual dependency - a1ien, pull/2119

      This also fix part of #2117

    • Update hash comment to changed XxHash64 usage - dcousens, pull/2124

      As introduced in #1221

    • use --openssl-legacy-provider flag when running node 17 - caass, pull/2116

      Closes #2108

      Node 17 uses openssl 3 by default, which is not compatible with webpack 4. This change adds a check to see if the user is running node 17, and if they are, we add the appropriate flag to the call to node (https://github.com/no ... truncated

v1.19.4

  • Features

    • Force flag for kv commands to avoid user confirmation prompt - andersquist, pull/2100

      Make it possible to override any user confirmation prompt when performing KV bulk deletes.

  • Fixes

    • Allow running wrangler dev without an account_id - jyn514, pull/2030

      Previously, running it with no account_id and no config in ~/.wrangler would give an error

    • Don't delete unused assets for sites - threepointone, pull/2096

      We have a consistency issue when after uploading a new site, it takes a while for the worker to propagate, so incoming requests will try to fetch older assets, and 404.

    • Allow host to be passed while authenticated - jspspike, pull/2091

      This also adds an --unauthenticated flag to dev and fixes bad error message

    • Fix wrangler dev session expiration - Electroid, pull/2071

      This PR fixes the issue described in #2068. API errors related to syntax or preview token return the same bad_request status code, and the changes should address these two particular cases.

    • Move TOML table fields come later than other fields - antiphoton, pull/2085

      Reorder manifest fields so that table fields (durable_objects, kv_namespaces, site) come later than string fields. The reason table fields need to come later is the same as #2037.

    • Only warn about compatibility dates on publish - Electroid, pull/2080

      Changes the compatibility date warning to only occur when using wrangler publish. We do not want this warning to be sent when using other commands like wrangler tail, etc.

v1.19.3

  • Features

    • Generate compatibility_date in wrangler init and warn if it's missing - kentonv, pull/2070

      We'd like everyone to specify a compatibility_date in their wrangler.toml going forward. This change both adds the date automatically to newly-generated wrangler.toml files, and warns when parsing old ones that are missing it.

      See ... truncated

    • Implement wrangler dev --inspect - jyn514, [pull/2016]

      Helps with #946.

      How do I use this?

      1. wrangler dev --inspect, which should have output similar to this:
      💁  watching "./"
👂  Listening on http://127.0.0.1:8787
[2021-07-28 13:38:20] GET worker.jnelson.workers.dev/ HTTP/
... truncated

[jyn514]: https://github.com/jyn514
[pull/2016]: https://github.com/cloudflare/wrangler/pull/2016

    • OAuth integration for Wrangler login - ocsfrank, pull/2048

      Still in staging, but it should be ready for review.

      Please also ignore the cargo.toml, I will update it once the service goes to prod. The CI test are failing because of oauth2 library.

    • Tagged Durable Objects Migrations - xortive, pull/1992

      fixes #1950

      This PR adds support for tagged migrations to wrangler. Migration tags are a way to ensure that you are applying the correct migration to a given script that contains a DO.

      They are also used for allowing multiple migrati ... truncated

  • Fixes

    • Allow showing a backtrace for errors returned from main - jyn514, [pull/2059]

      Example usage:

      $ RUST_BACKTRACE=1 wrangler dev --inspect
Error: missing field `prewarm` at line 1 column 627

Stack backtrace:
   0: <core::result::Result<T,F> as core::ops::try_trait::FromResidual<core::result::Result<core::con
... truncated

[jyn514]: https://github.com/jyn514
[pull/2059]: https://github.com/cloudflare/wrangler/pull/2059

    • Fix wrangler dev session expiration - ocsfrank, pull/2071

      This PR fixes the issue described in #2068. API errors related to syntax or preview token return the same bad_request status code, and the changes should address these two particular cases.

      In the future, it would be great to update th ... truncated

    • fix: remove panic when send fails, log error instead - nilslice, pull/2061

      Removes the panic on send failure from the last remaining place for this particular --watch error.

v1.19.2

  • Maintenance

    • fix: override rust type to javascript - nilslice, pull/2055

      the rust type seems to only have been added as a way to automatically trigger wasm-pack, but it's much better to put that in a custom [build] step instead. eventually, we really should phase out the rust type from wrangler altogether.

      this is a side-effect of releasing the new Rust template https://github.com/cloudflare/rustwasm-worker-template

    • update crates.io badge in the readme - jim4067, pull/2052

v1.19.1

  • Fixes

    • Allow worker with no routes to be published - jspspike, pull/2024

      Fixes #1700

    • Changed devtools reconnect warning message to log info - jspspike, pull/2027

      Changed devtools reconnect warning message to log info

    • dont eat errors - caass, pull/2022

      Print the original error when we create an error report.

      This should help in debugging install errors, and other stuff where we cant report it due to wrangler not being installed correctly

    • Downgrade panic when no account IDs match a token to an error - jyn514, pull/2042

      I'm still not sure exactly how this could happen, and I feel there's a bug here somewhere ... but people are hitting this in practice (enough that someone ran wrangler report) and this will help the user experience in the meantime.

      ... truncated

    • Load account ID before uploading cron triggers - jyn514, pull/2023

      Fixes #2021.

    • Log bad request error details during dev - mrbbot, pull/2041

      Currently, when the edge preview service returns a 400 Bad Request error during a wrangler dev session because the user's worker is invalid, all you see is Error: HTTP status client error (400 Bad Request) for url. When calling `wrangle ... truncated

    • Remove build.rs - jyn514, pull/2029

      While working on tests for get_deployments() in #2023, I found that any change to src/settings/toml/tests/mod.rs would cause the whole Wrangler library to be rebuilt, not just the tests. I fo ... truncated

    • Small typo in error message - robertaboukhalil, pull/2031

      I came across this error message:

      Please specify your deployment routes or wrangler_dev = true inside of your configuration file

      But I think wrangler_dev should be workers_dev

  • Maintenance

    • 2032: Upgrade tar dependency - dhaynespls, pull/2033

      Fixes #2032

    • Add cargo audit and npm audit to CI - jyn514, pull/2034

      This runs:

      • On PRs that modify Cargo.toml or Cargo.lock
      • weekly

      and opens issues if either cargo or npm report an error. This does have false positives occasionally, especially for dev-dependencies, but catching vulnerabilities seem ... truncated

    • Bump assert_cmd from 1.0.7 to 2.0.0 - dependabot, pull/2019

v1.19.0

  • Features

  • wrangler tail using WebSockets - nataliescottdavidson, pull/2005

    This change removed dependency on cloudflared binary, and increased tail script RPS limit to 100!

v1.18.0

  • Features

    • feat: add test for kv sync for site projects - nilslice, pull/1996

      Adds test for quickfix made in #1976.

      TODO: consider adding types for the Key / Value pairs used in this code. It's a bit confusing to rely the comparison of String types which have been considerably modified from the assumed original ... truncated

    • feat: add usage_model to metadata - taylorlee, pull/1961

      and remove toggling usage model after upload this shouldn't be merged until the upload api supports usage_model in metadata

    • Don't require an account_id in wrangler.toml - jyn514, pull/1966

      Helps with #331. I recommend reviewing this commit-by-commit.

      Note that unlike the suggestion in the original issue, this actually never writes to the wrangler.toml at all, which will hopefull ... truncated

  • Fixes

    • fix: improve error message when no deployment info is provided - nilslice, pull/1997

      Fixes #1694.

      Adds clarification and links to documentation on how the project must be configured in order to deploy it.

    • Fixed issue with tail not deserailzing event properly - jspspike, pull/1994

      Also added color to wrangler dev and wrangler tail -f pretty json outputs if deserialized properly Fixes #1990

    • Temporary workaround to avoid dropping a tokio::runtime while another is running - jyn514, [pull/1991]

      This fixes the following error from wrangler tail:

      $ wrangler tail
🦚  Setting up log streaming from Worker script "test-project". Using ports 8080 and 8081.
This may take a few seconds...

Oops! wrangler encountered an error.
... truncated

[jyn514]: https://github.com/jyn514
[pull/1991]: https://github.com/cloudflare/wrangler/pull/1991

    • Made config add target_type to type instead of "target_type" - jspspike, pull/1988

      Fixes #1960

      We used serde to rename the target_type field to type but when making the toml ourselves we still used target_type

    • fix: use asset manifest for sync op - nilslice, pull/1976

      After #1970, the vec of assets used in the sync operation no longer included the assets that were already uploaded. this diff of assets was re-used to determine remote assets to delete, which would in turn remove files that should have re ... truncated

    • Remove misleading rustup install and update rust-toolchain - jyn514, pull/1968

      Previously, CI ran rustup install stable just before running lints, but didn't actually use the new toolchain. This removes the unneeded download and updates rust-toolchain to the latest stable version.

      This also fixes a bunch of cli ... truncated

  • Maintenance

    • Cleanup code in socket.rs - jyn514, pull/2000

      This can just use ? instead of an explicit match.

    • Tell Dependabot to only make PRs for new major versions - jyn514, pull/1987

      This avoids constant small PRs which aren't necessary. Dependabot will still make PRs for security updates, which is good because we keep Cargo.lock checked into git and Cargo will keep using yanked versions until you update Cargo.lock.

      ... truncated

v1.17.0

  • Features

    • feat: capture panics and generate error report - nilslice, pull/1888

      This PR adds support for wrangler error reporting. Currently, all panics are caught and a report is generated, then written to disk. wrangler report or wrangler report --log <file-name.log> will upload the corresponding error report to Cloudflare

  • Fixes

  • Maintenance

🍏 v1.16.1

  • Features

    • Add wasm_modules config field for bundling arbitrary WebAssembly modules. - losfair, pull/1803

      Currently it seems that wrangler only supports WebAssembly modules included from a rust or webpack project.

      This pull request enables the inclusion of arbitrary WebAssembly modules through a wasm_modules config field.

  • Fixes

    • fix: use x86_64 arch for pre-built downloads on M1 devices - nilslice, pull/1876

      This PR forces the use of a pre-built x86_64 binary on a aarch64/arm64 Apple system. For M1 devices specifically, it will fix wrangler generate, and wrangler build for type = "rust" wrangler projects.

      There is another semi-related ... truncated

    • chore: include @cloudflare/binary-install directly as source - nilslice, pull/1877

      This PR inlines the @cloudflare/binary-install dependency as a quickfix for #1811, in which usage reports indicated that the module occasionally failed to install.

      I tested this by running npm i && node run-wrangler.js on both `npm ... truncated

    • Stop assuming KV values are UTF8 text - koeninger, pull/1878

      Implementation of kv:key get is calling text() on the result of the api call, so it's replacing non-utf8 values with ef bf bd, the unicode REPLACEMENT CHAR. KV values can be arbitrary bytes, we shouldn't assume they're UTF8 text, so this p ... truncated

1.16.0

  • Features

    • Custom Builds and Modules - xortive, caass, pull/1855

      Custom Builds and Modules has finally landed in a main release! There's too many new features to write about in a changelog, so here's a link to the docs.

    • add --format option, including default json and new pretty - caass, pull/1851

      You can now pass --format pretty to wrangler tail to get pretty printed logs! --format json is also available, and gives you the existing JSON-formatted logs.

  • Fixes

    • Revert "Print line and column numbers for exception thrown (#1645)" - Electroid, pull/1835

      This reverts commit 74a89f7c383bc22758cbe55096ce3016c5c319d7.

      Closes #1826

      This commit is causing wrangler dev to not show uncaught exceptions. Reverting chrome-devtools-rs was also necessary.

      We have a fix in progress to fix the underlying issue and re-introduce line and column numbers.

    • Don't generate usage_model = "" by default - xortive, issues/1850

      Generating usage_model = "" by default was violating the toml spec, which broke wrangler init --site as usage_model came after [site].

1.15.1

  • Features

  • Fixes

    • fix: remove unused import of WasmMainTemplatePlugin - jasikpark, pull/1802

      This should improve #1721. #1721 (comment)

    • Hot fix for error message helper not working - Electroid, pull/1847

      The JSON is pretty printed, which means it contains a space.

    • Revert "Print line and column numbers for exception thrown (#1645)" - Electroid, pull/1835

      This reverts commit 74a89f7c383bc22758cbe55096ce3016c5c319d7.

      Closes #1826

      This commit is causing wrangler dev to not show uncaught exceptions. Reverting chrome-devtools-rs was also necessary.

1.15.0

  • Fixes

    • fix: remove unused import of WasmMainTemplatePlugin - jasikpark, pull/1802

      This should improve #1721. #1721 (comment)

    • Revert pull/1748 - xortive, pull/1804

      pull/1748 turned out to interact poorly with workers-sites projects, causing npm install to not be run, breaking the build.

      We're going to revert this change. Folks already depending on it should use custom builds (pull/1677) once it makes it into a release candidate.

      We incremented the minor version number instead of making a patch release, as this change is more significant than a simple bugfix.

1.14.1

  • Fixes

    • revert default install location change - xortive, pull/1798

      In 1.14.0, we changed the default install location from ~/.wrangler to node_modules, to allow npx wrangler to use the pinned version in a project's dependencies. It seems that this is causing issues, so we're rolling it back in favor of having folks who want this behavior, to specify a the install location in the config section of package.json. We'll document this soon.

1.14.0

  • Features

    • Display account ID's when possible - caass, pull/1786

      Previously, users had to log in to their dash to view their account IDs. This PR extracts some of the code used in whoami to log their account IDs to the terminal, instead. If for some reason that doesn't work, it'll fall back to telling them to look at the dashboard

    • Feature/monorepo support - thefill, pull/1748

      Improvement of the detection of node_modules dir - adds bottom-up traversing to find it in parent dirs.

      This should resolve some issues experienced with monorepos (e.g. with nx).

    • fix installer - xortive, pull/1780

      binary-install@0.1.1 broke semver, this PR changes our installer to use a cloudflare-owned version of binary-install with updated dependencies to resolve the previous vulnerability warnings from the old version of axios that was being used.

      This is also a feature, since we now install the wrangler binary in node_modules instead of ~/.wrangler. This means installing wrangler as a dev dependency works as expected -- npx wrangler will run the version in your dev dependencies, not the globally installed wrangler.

      Wrangler will now also support setting the WRANGLER_INSTALL_PATH environment variable to choose where you install the wrangler binary. This environment variable must be set when running wrangler, as well as when installing it.

    • kv put with metadata - ags799, pull/1740

      Closes #1734

  • Fixes

    • don't panic on Client build - ags799, pull/1750

      This won't fix the issue #1743 but it should give us some more context.

    • endlessly retry connection to devtools websocket - ags799, pull/1752

      Endlessly retry connection to preview's devtools websocket on wrangler dev. With exponential backoff.

      Keeps us from panicking in issue/1510.

    • fix console formatting - mdycz, pull/1749

      Fixes #1707

  • Maintenance

    • Bump futures-util from 0.3.11 to 0.3.13 - dependabot, pull/1778

    • Remove extra required_override call from sites ignore logic - xortive, pull/1754

    • remove webpack specific change - xtuc, pull/1730

      We used to hook directly into webpack internals to rewrite the runtime that loads Wasm to make it point to the Worker binding instead of a network fetch.

      This change removes the webpack specific change and injects a generic runtime to ... truncated

    • Set panic to abort in release mode - ObsidianMinor, pull/1762

      This should fix cases where we spawn threads that panic and get us in an invalid state that requires us to get killed anyway.

    • Tweak issue templates - Electroid, pull/1776

      Made minor edits to the issue templates

    • update binary-install to avoid vulnerable axios version - simonhaenisch, pull/1726

      Fixes the security warnings when installing the wrangler npm package!

    • Update README.md for windows install - koeninger, pull/1779

      note about #1765

    • Update release.yml - xortive, pull/1783

      thanks @rajatsharma for spotting this :)

1.13.0

  • Features

    • Add support for text blob bindings - xortive, pull/1543, issue/483

      Wrangler now supports text blobs! Text blobs are values to use in your workers, but are read from a file instead of a string in your TOML.

      Usage:

      text_blobs = { FOO = "path/to/foo.txt", BAR = "path/to/bar.txt" }

  • Fixes

  • Maintenance

❗️ 1.12.3

  • Fixes

    • Bump OpenSSL version with vulnerability patch - pull/1684

❗️ 1.12.2

❗️ 1.12.1

⏰ 1.12.0

🌊 1.11.0

  • Features

    • New configuration method with wrangler login - jspspike, pull/1471

      wrangler login allows you to authenticate Wrangler to use your Cloudflare user credentials without hunting down an API token in the Cloudflare dashboard. It's straightforward! Just run wrangler login, enter your credentials, and you're off to the races.

    • wrangler dev now runs on the same machines as your production code - EverlastingBugstopper avidal jwheels, pull/1085

      When running wrangler dev as an authenticated user, your requests will now run on the same servers that Cloudflare Workers run on in production. This means that what you see is what you get. wrangler dev should behave exactly like production, though we still recommend deploying to a staging website before going to production in order to ensure your changes are safe. This change means you get access to things like request.cf, the Cache API, and any Cloudflare settings you've applied in the dashboard while developing your Workers.

    • wrangler dev now supports redirects` - jspspike, issue/1508 pull/1512

      When an HTTP response status code is between 300 and 399, the server lets the client know that the data they're looking for isn't here anymore, and the client should issue another separate request to the endpoint denoted in the Location header of the response. Before, if you were developing with wrangler dev on example.com, and your Worker issued a redirect from https://example.com/oldurl to https://example.com/newurl, that's what would be in the Location header. This meant that whatever client you were using would then issue their second request to the public Internet rather than the wrangler dev server running on your local machine. With this release, the Location header would now be rewritten to http://127.0.0.1:8787/newurl, preventing your client from redirecting you away from the Worker you're trying to develop.

    • Add --config flag to all commmands to override the default wrangler.toml configuration file path - luanraithz, issue/1064 pull/1350

      All commands that read from a configuration file can now use a different configuration file path if the --config flag is specified. The commands affected are: kv:namespace, kv:key, kv:bulk, route, secret, build, preview, dev, publish, subdomain, and tail.

    • wrangler dev configuration options for HTTP protocol - jspspike, issue/1204 pull/1485

      wrangler dev now takes two additional configuration flags: --upstream-protocol and --local-protocol. Both of these take a value of either http or https. --upstream-protocol determines what protocol the request going to your preview worker is (previously this was only controlled with the --host flag) - this flag defaults to https. --local-protocol determines what protocol wrangler dev listens for and defaults to http. If https is chosen, a self-signed certificate will be auto-generated for the dev server.

    • wrangler dev can be configured in wrangler.toml - jspspike, issue/1282 pull/1477

      Any flag taken by wrangler dev (except --host) can now be configured in the [dev] section of your wrangler.toml. This allows different developers on the same project to share and persist settings for their local development environment.

    • Check if rustc is installed before building a Rust project - ObsidianMinor, issue/487 pull/1461

    • Improve preview_id error message - EverlastingBugstopper, issue/1458 pull/1465

      When a preview_id is needed, the error message directs the user to add it to their wrangler.toml.

    • Prevent wrangler preview and wrangler dev for a Workers Site if there is no authenticated user - jspspike, issue/1138 pull/1447

  • Fixes

    • Fix wrangler route commands that take an --env flag - jspspike, issue/1216 pull/1448

      Before, if you passed an environment to a wrangler route command, it wouldn't work properly due to some mishandling of the arguments in the way we used our command line argument parser. This is now fixed and wrangler route commands work as expected.

    • Open browser as child process - jspspike, issue/516 pull/1495

      When running wrangler preview, the browser is now opened as a child process. This fixes an issue on Linux where Wrangler would start the browser and then hang waiting for the browser to exit before it begins watching for changes.

    • Direct cloudflared output with wrangler tail to /dev/null - jspspike, issue/1432 pull/1450

  • Maintenance

🐹 1.10.3

  • Features

    • wrangler dev listens on IPv4 by default - EverlastingBugstopper, issue/1198 pull/1405

      Before, wrangler dev would listen on [::1]:8787 by default, and call it localhost in the terminal output. This was confusing for developers whose localhost resolves to IPv4 and not IPv6. Now, wrangler dev will listen on 127.0.0.1:8787 by default. This can be overriden by passing values via the --ip and --port flags.

    • Clarify where to find your account_id in the dashboard - EverlastingBugstopper, issue/1364 pull/1395

      When you create a new project with wrangler generate, it directs you to the Cloudflare Dashboard to find your account_id and zone_id. However, this flow only worked if you had your own domain. Developers who only use workers.dev for their Workers were directed to a page that does not exist! This message now points everyone to a page where they can find the information that they need.

  • Fixes

    • Allow creation of preview namespaces when a namespace already exists in wrangler.toml - EverlastingBugstopper, pull/1414

      When we introduced KV preview namespaces, we made sure to add nice messages when creating new namespaces so people could easily add the new namespace id to their wrangler.toml in the correct place.

      However, we missed a very common case where developers already have a production namespace defined in their wrangler.toml and they want to add a preview namespace. When this is the case, we returned an error message intended to only be thrown when running either wrangler preview or wrangler dev. This is now fixed!

    • Allow multiple response header values in wrangler dev - EverlastingBugstopper, issue/1412 pull/1413

      Before, wrangler dev would not properly handle response headers that have multiple values. We would iterate over all response headers coming from the Workers runtime, and "insert" them into the header map instead of appending them. This is no longer the case and response headers should now work as expected. More details on this issue can be found here.

    • Fix kv-namespace/kv_namespace behavior in environments - EverlastingBugstopper, issue/1408 pull/1409

      When KV namespace support was initially added to Wrangler, we documented using kv-namespaces in wrangler.toml. Unfortunately, the - was not consistent with other fields such as zone_id and account_id, so the decision was made to allow both kv-namespaces and kv_namespaces. When this change was introduced, it worked with top level kv_namespaces entries, but not in environments. This is now fixed! You can now use kv_namespaces everywhere you can use kv-namespaces.

  • Maintenance

    • Add link to testing docs in CONTRIBUTING.md - luanraithz, pull/1416

    • Remove unused Krate::install code - EverlastingBugstopper, [issue/247] pull/1410

      When we introduced our own version checking for Wrangler we stopped using Krate::install. This PR just removes that unused code.

💪 1.10.2

  • Fixes

    • Fixes bug preventing wrangler publish from deleting over 10000 keys - ashleymichal, issue/1398 pull/1400

      In 1.10.0 we removed the limit on bulk KV PUT and DELETE operations by batching the values but missed batched deletes for Workers Sites. Now with some refactoring all batched operations use the same logic and have the same behavior.

⏳ 1.10.1

  • Fixes

    • reinstate longer timeout on bulk uploads for sites - ashleymichal, pull/1391

      In 1.10.0 we introduced a bug that reduced the timeout for bulk uploads back to the standard 30 seconds. This fixes that and restores the five minute bulk upload/delete timeout.

    • Increase default timeout to one minute - EverlastingBugstopper, pull/1392

      For folks with slower connections we're increasing the timeout for API requests.

♻️ 1.10.0

  • Features

    • wrangler dev now requires that you specify "preview" versions of KV namespaces - EverlastingBugstopper, ashleymichal, issue/1032 pull/1357 pull/1359 pull/1360

      In order to prevent you from accidentally stomping on production data in Workers KV, we're introducing the concept of explicit preview namespaces. When running wrangler dev, if you're using Workers KV, you'll need to specify a specific KV Namespace to use when previewing the Worker.

      Specifically, this change:

      • Adds a preview_id field to items in kv_namespaces in wrangler.toml that must be provided in order to preview a worker that has kv namespaces.
      • also adds --preview to kv commands in order to interact with them instead of production namespaces.

      If you define a KV Namespace in your wrangler.toml but don't specify a preview_id, and then try to run wrangler dev, you'll see the following:

      $ wrangler preview
💁  JavaScript project found. Skipping unnecessary build!
Error: In order to preview a worker with KV namespaces, you must designate a preview_id for each KV namespace you'd like to preview.

      More details can be found in the documentation.

    • wrangler subdomain allows you to rename your existing workers.dev subdomain - nprogers, issue/1279 pull/1353

      Workers recently introduced the ability to rename your workers.dev subdomain in the Workers dashboard. This change allows you to do the same thing from wrangler.

    • Return more detailed error when preview uploads fail - jahands, issue/1330 pull/1356

      When running wrangler dev or wrangler preview, if the upload failed for validation or other reasons, we displayed pretty obtuse errors. Now we pass along more informative error details.

    • Check for wrangler updates every 24 hours and message if you should update - EverlastingBugstopper, jspspike, issue/397 pull/1190 pull/1331

      wrangler will now let you know if there's an update available, and will only bug you once every 24 hours.

    • Improve some error messages with human-readable suggestions - jspspike, issue/1157 pull/1329

      Previously, you would see not-super-helpful error messages if your API Token was expired or missing some permissions, didn't have Workers Unlimited enabled and tried to upload to KV, or tried to create a namespace that already existed. But we strive for helpful, informative error messages!

      Now, you'll see the following error messages as appropriate:

      10026 => "You will need to enable Workers Unlimited for your account before you can use this feature.",
10014 => "Namespace already exists, try using a different namespace.",
10037 => "Edit your API Token to have correct permissions, or use the 'Edit Cloudflare Workers' API Token template.",

  • Fixes

    • wrangler dev pretty-prints JSON messages from console.log() - jspspike, issue/1249 pull/1371

      Previously, when running wrangler dev, you could console.log(JSON.stringify(some_object)), but the output was hard to read: newlines weren't appropriately parsed and whitespace was a bit of a mess.

      This change pretty-prints JSON received from the Inspector, so JSON output is much easier to read in wrangler dev.

    • Don't install wasm-pack when publishing a type:webpack project - EverlastingBugstopper, issue/745 pull/1344

      We shouldn't install wasm-pack if your project doesn't need it. We thought we fixed this in 1.7.0, but we didn't. This time it's fixed for real, we swear.

    • Pin wasm-pack and cargo-generate to specific versions instead of always downloading the latest version - jspspike, issue/1240 pull/1358

      Previously, every time wasm-pack or cargo-generate was needed, Wrangler would download and install the latest version. Now, we pin those dependencies to a specific version in every release of Wrangler. If you have either of these tools installed locally, Wrangler will use them if you have a more recent version.

    • Support batched KV PUT and DELETE operations to improve Workers Site publish performance - jspspike, issue/1191 pull/1339

      Previously, you couldn't use Workers Sites if you wanted to upload more than 10,000 static files, and it took a lot of time to upload close to that many files. This change instead batches upload and delete calls, allowing us to increase the limit and improve performance for everyone.

  • Maintenance

    • Add Code of Conduct - EverlastingBugstopper, pull/1346

      We actually didn't have one before, so now we do!

    • Deprecate undocumented KV bucket attribute - ashleymichal, issue/1136 pull/1355

      The bucket attribute was not originally intended to be released into the wild. We announced deprecation in March, and are now removing it as an available configuration property.

    • Refactor to separate KV commands, implementation of said commands, and site-specific logic - ashleymichal, pull/1332

      This refactor makes the logic for interacting with KV more re-usable and maintainable.

    • Add SECURITY.md with responsible reporting guidelines - EverlastingBugstopper, pull/1345

🐼 1.9.2

  • Fixes

    • Fix piping secret values to wrangler secret put <VAR_NAME> - dmcgowan, issue/1322 pull/1316

      In 1.9.1, we introduced a bug where piping values to wrangler secret put no longer worked. In 1.9.2, that bug is squashed, and the command works as expected.

🐎 1.9.1

  • Features

    • Accept --verbose for every command - bradyjoslin, issue/975 pull/1110

      Not every command outputs additional information when you pass --verbose, but none of them will fail to run if you pass --verbose after this change.

    • wrangler dev checks if the specified port is available - EverlastingBugstopper, issue/1122 pull/1272

      When starting up wrangler dev, it now checks to see if the requested port is already in use and returns a helpful error message if that's the case.

  • Fixes

    • Don't reinstall vendored binaries on every build - EverlastingBugstopper, issue/768 pull/1003

      You may have noticed some very verbose and over-eager installation output when running Wrangler. Every webpack type build would install wranglerjs and wasm-pack. This was... super annoying and not a great experience, especially when running wrangler preview --watch or wrangler dev. Each time you'd change a file, Wrangler would reinstall those external dependencies. This doesn't happen anymore! Wrangler will still download and install these external dependencies, but only if you have an outdated version.

    • Remove redundant builds - EverlastingBugstopper, issue/1219 pull/1269

      When running wrangler preview --watch or wrangler dev on a webpack type project, Wrangler will provide a new build artifact and upload it via the Cloudflare API. Before, we'd start a long-running webpack --watch command, in addition to running webpack on every change. We were running two builds on every change! This was not great and has been removed. This, combined with the above fix removing redundant installations, should greatly improve your dev iteration cycles.

    • wrangler dev will reconnect to the devtools WebSocket after being disconnected - EverlastingBugstopper, issue/1241 pull/1276

      wrangler dev initiates a WebSocket connection via the Cloudflare API in order to stream console.log messages to your terminal. Over time, it's very likely that the WebSocket would be disconnected. When this happened, Wrangler would panic, requiring developers to restart the process. Now, if wrangler dev gets disconnected, it will issue a reconnect request, allowing developers to run wrangler dev as long as they are connected to the Internet.

  • Maintenance

    • Adds Developing Wrangler section to CONTRIBUTING.md - EverlastingBugstopper, issue/270 pull/1288

      We love external contributors, and what better way to help get folks kickstarted than to add some documentation on developing Wrangler? Check out CONTRIBUTING.md if you're interested in helping out.

    • Remove --release from wrangler publish --help - EverlastingBugstopper, pull/1289

      We deprecated wrangler publish --release a long time ago in favor of environments, but it's still an accepted argument to preserve backwards compatibility. Now, it no longer shows up in wrangler publish --help as an accepted argument, even though it's still an alias of wrangler publish.

    • Updates license file to wrangler@cloudflare.com - EverlastingBugstopper, pull/1290

      The copyright in our MIT license was outdated and pointed to the email address of @ashleygwilliams (who no longer works at Cloudflare 😢). Now it points to wrangler@cloudflare.com :)

    • Add Dependabot to Wrangler - ispivey, pull/1294

      Wrangler now uses Dependabot to automatically update dependencies. We had already been doing this on a sort of ad-hoc basis, but this should make it much easier to stay on top of updates!

    • Remove unused code warnings - ashleymichal, pull/1304

      For our integration tests we create fixtures containing sample Workers projects. When we ran cargo test, cargo would say that the code used to create said fixtures were unused (which was not true). This PR moves the fixture code to a place where cargo says "This is a Fine Place for This Code."

    • Handle clippy warnings - ashleymichal EverlastingBugstopper, pull/1305 pull/1306

      cargo clippy is a helpful little tool that helps you write more idiomatic Rust. Over time, we've developed an immunity to the warnings produced by this tool, and we took a stab at cleaning some of them up.

🦚 1.9.0

  • Features

    • Wrangler Tail - ashleymichal, gabbifish, EverlastingBugstopper, pull/1182

      Wrangler Tail introduces a way to view console statements and exceptions live as they occur in your Worker. Simply run wrangler tail against any deployed Worker and pipe the output through jq or to a file to stream trace events per request.

    • Much faster build times for Workers Sites projects - EverlastingBugstopper, pull/1221

      When you deploy a Workers Site, Wrangler generates a unique hash for each file. It does this so that your Worker does not serve stale files from Cloudflare's edge cache to end users. Unfortunately, generating these hashes took a really really long time since we were using a cryptographically strong hash. Since we're just using this hash for cache invalidation, we decided it's not necessary to use such a complicated algorithm. We switched to using xxhash and have seen noticeable speed improvements.

    • Add --url to wrangler preview - larkin-nz, issue/351 pull/1001

      wrangler preview now has the ability to open any URL! Before, wrangler preview would always open example.com, and you would be able to change the URL in the browser only. Now, you can use wrangler preview --url https://mysite.com/my-amazing-endpoint and your preview session will get started off on the right foot.

    • Print email addresses for API token users on wrangler whoami - dhaynespls, issue/863 pull/1212

      Before, if you ran wrangler whoami as an API token user, you didn't get much info. Due to some heavy lifting by folks working on the Cloudflare API, API token users with the correct permissions can now see what email address they are authenticated with when they run wrangler whoami. Nifty!

    • wrangler generate auto increments default worker name - xprazak2, issue/58 pull/469

      When wrangler generate is run without a name for the worker, it will find a name for the worker that does not already exist in that directory.

    • Standardize colors in stdout - EverlastingBugstopper, pull/1248

      Wrangler likes to print colors where appropriate, and now there is a standard module for printing different colors that is used across the codebase.

    • Suggests wrangler init if wrangler.toml does not exist - ashleymichal, issue/827 pull/1239

      When there is no wrangler.toml in a directory you're trying to run Wrangler in, it doesn't know what to do. The way to fix this is to make a wrangler.toml, and the way to do that is to run wrangler init.

  • Fixes

    • Allow kv-namespaces and kv_namespaces - EverlastingBugstopper, issue/1158 pull/1169

      Most fields defined in wrangler.toml are one word, but some of them are two! In the past, we usually use _ to separate words, but somehow we used a - for kv-namespaces. This was inconsistent and a bit confusing. Now we allow both for the sake of backwards compatibility, but in the future we'll try to stick to snake_case.

    • Typo fix in wrangler init - jplhomer, pull/1210

      A successful wrangler init execution used to output "Succesfully" instead of "Successfully", but not anymore!

    • More granular errors in wrangler dev - EverlastingBugstopper, pull/1251

      In the last release we added an error message in wrangler dev for failed uploads. Unfortunately it was a bit overeager and some information about different types of errors were lost. This behavior has been fixed!

  • Maintenance

    • Unify wrangler's user agent - EverlastingBugstopper, issue/731 pull/1070

      Wrangler sure does send a lot of API requests! Before, about half of the API requests Wrangler sent would send them with the HTTP header User-Agent: wrangler. Now, all requests sent by Wrangler include that User Agent. This lets the APIs we use know that the request is coming from this tool. Yay for being good netizens!

    • Refactors and documentation of wrangler dev - EverlastingBugstopper, pull/1220

      No behavior changes with this one, just some improvements to code layout and some extra documentation comments. Check it out if you're interested!

🎭 1.8.4

  • Fixes

    • Don't remove user configuration on npm installs - EverlastingBugstopper, issue/1180 pull/1181

      Wrangler started removing user's authentication configuration files on reinstallation from npm - this is no good and is fixed in this release.

    • Allow multiline files to be piped to wrangler secret put - EverlastingBugstopper, issue/1132 pull/1171

      Previously, if you tried to pipe a multiline file to wrangler secret put, the secret would only upload the first line of the file. This... was not helpful - cat hello_world.txt | wrangler secret put should behave as expected with this release.

  • Maintenance

    • Bump GitHub Actions checkout version - imbsky, pull/1170

      GitHub Actions are pretty nifty, and we've started using them as our CI provider in Wrangler. Actions allow you to specify a step that "uses" a template, and one of the most used templates is the template that checks out relevant code. GitHub just released v2 of that template, and our CI now uses it!

🍟 1.8.3

  • Features

    • Improvements to the Workers Sites asset manifest - EverlastingBugstopper, issue/897 pull/1145

      Workers Sites uses the concept of an asset manifest to invalidate Cloudflare's cache when new files are published. Every time you publish your Workers Site, Wrangler will re-create a mapping of file names with a hash to the contents of the file. This release includes a few steps that improve this experience:

      • Manifest sizes are smaller by a magnitude of ~6.4. This should help some folks who were previously running into size issues when uploading a Workers Site.

      • Any time an asset manifest is created, you will see the files that are being hashed in real time with a fancy loading spinner - no more waiting without any information!

      • Asset manifest creation is now faster due to a refactor.

    • Clarify mutual exclusivity of zoneless v. zoned deploys - EverlastingBugstopper, issue/1152 pull/1154

      When publishing a Worker, you must specify either workers_dev = true or both a zone_id and route/routes. Previously, if your wrangler.toml violated this requirement, it would error with the following message:

      $ wrangler publish
Error: you must set workers_dev = true OR provide a zone_id and route/routes.

      It's technically correct, but we can make it even more clear what the issue is. The new error message looks like:

      $ wrangler publish
Error: you must set EITHER workers_dev = true OR provide a zone_id and route/routes.

  • Fixes

    • Fixes wrangler config information message - EverlastingBugstopper, pull/1164

      In Wrangler 1.8.2, we updated the formatting of some of Wrangler's informational messages. Unfortunately when this was introduced, it came with a bug in wrangler config that made the message read out in the wrong order. This is fixed in this release!

  • Maintenance

    • Remove unused badges from README - EverlastingBugstopper, pull/1166

      We no longer use Azure Pipelines as our CI provider, nor do we run non-test builds in CI so we removed those badges from the README.

🐈 1.8.2

  • Features

    • Configurable binary host URL - noherczeg, pull/1018

      Previously, binaries installed by Wrangler were all assumed to come from npm. If you work in a controlled environment and can only install binaries from a specific endpoint (instead of npm), you can now specify that endpoint using the WRANGLER_BINARY_HOST environment variable.

  • Fixes

    • Eliminate downtime when redeploying Workers Sites - ashleymichal, issue/783, pull/1115

      When Workers Sites were first introduced, redeploying a site could lead to a few seconds of downtime if the Worker upload fails. Specifically, if a new Workers Sites upload failed, it was possible that the old, now-unused files in Workers KV would be deleted anyways, meaning that the preexisting Workers Site would suddenly have missing resources. This fix waits to delete now-unused files until after a new Workers Sites script is published.

  • Maintenance

    • Add npm badge to README - tomByrer, [pull/1121]

      Add badge to README that points to npm page for Wrangler.

    • Unify attention-grabbing messages - EverlastingBugstopper, [pull/1128]

      Use more actionable, easy-to-read information printouts throughout Wrangler.

😈 1.8.1

  • Features

    • Error messaging for internet required to talk to Cloudflare API - EverlastingBugstopper, issue/1093 pull/1114

      With the release of wrangler dev in 1.8.0, it was not clear to users that internet is required since the feature communicates with Cloudflare's API. With this error message, users without internet connection are shown actionable next steps - check internet connection and lastly check if Cloudflare's API is down.

  • Fixes

    • Fix live reload for wrangler dev - EverlastingBugstopper, issue/1082 pull/1117

      wrangler dev re-builds and re-uploads your script to the Cloudflare API when it detects a file change. The Cloudflare API returns a new token which allows wrangler dev to route subsequent requests to the new script. Previously, wrangler dev would re-build, re-upload, and receive the new token, but it wouldn't use it for a couple of minutes due to some faulty threading logic. (darn mutexes!) After this change, wrangler dev will block incoming requests when it is switching the token, thus fixing the issue.

    • Remove unneeded carriage return in wrangler secret put - gabbifish, issue/1109 pull/1112

      Previously, interactive input from wrangler secret put added a carriage return to the secret key/value pairs on Windows. This no longer happens and input is parsed properly before uploading.

🙊 1.8.0

  • Features

    • wrangler dev - EverlastingBugstopper, issue/845 pull/883

      wrangler dev is a local proxy server to Cloudflare's preview service, allowing you to automatically re-build and preview your application on localhost. This feature is in alpha and we're looking for feedback and bug reports: check out this issue!

      wrangler dev works very similarly to wrangler preview, but instead of opening your browser to preview your Worker, it will start a server on localhost that will execute your Worker on incoming HTTP requests:

      $ wrangler dev

      You should be able to send HTTP requests to localhost:8787, along with any headers or other request data, and your Worker should execute as expected. Additionally, you'll see console.log messages and exceptions appearing in your terminal (!!!).

      For more information on wrangler dev's options, such as passing a custom host, ip, or port, run wrangler dev in your terminal for the available flags and options.

    • Multi-route support - ashleymichal, issue/866 pull/916

      Wrangler now allows developers to publish their Workers projects to multiple routes on a Cloudflare zone.

      To deploy your Workers project to multiple routes, you can migrate from the route key to routes:

      name = "worker"
type = "javascript"
account_id = "youraccountid"
# change this line
# route = "example.com/foo/*"
# to this line
routes = ["example.com/foo/*", "example.com/bar/*"]
zone_id = "yourzoneid"

    • wrangler secret commands - ashleymichal, bradyjoslin, issue/907 issue/909 issue/912 pull/1045

      Wrangler now allows developers to use secrets in their Workers codebase. Secrets are secure values that can be accessed as constants, similar to text variables, inside of your Workers code.

      To set a secret, you can use wrangler secret put MY_SECRET_NAME. The interactive prompt will ask for the secret text you'd like to add to your project:

      $ wrangler secret put MY_SECRET_NAME
Enter the secret text you'd like assigned to the variable MY_SECRET_NAME on the script named my-project

      Importantly, secrets are constrained to an environment, and do not carry over between different deployed Workers (e.g. my-worker and my-worker-production). This allows you to use different API keys, URLs, and other common "environment variable"-style values in your different environments. Specifying an environment can be done using the --env (or -e, for short):

      $ wrangler secret put MY_SECRET_NAME --env production
Enter the secret text you'd like assigned to the variable MY_SECRET_NAME on the script named my-project-production

      The wrangler secret subcommand also allows developers to list and delete secrets for your Workers projects:

      $ wrangler secret delete MY_SECRET_NAME
Are you sure you want to permanently delete the variable MY_SECRET_NAME on the script named my-project [y/n] y
🌀  Deleting the secret MY_SECRET_NAME on script my-project.
✨  You've deleted the secret MY_SECRET_NAME.

$ wrangler secret list
[{"name":"API_KEY","type":"secret_text"},{"name":"MY_OTHER_SECRET","type":"secret_text"}]

    • Plain text binding support - EverlastingBugstopper - issue/993 pull/1014

      In addition to secrets, Wrangler now also supports setting "plain text" bindings – values that will be available as constants in your Workers code, but aren't encrypted. This can be done by passing values in wrangler.toml under the vars key:

      name = "worker"
type = "javascript"
account_id = "your-account-id"
workers_dev = true
vars = { ENV = "staging" }
[env.prod]
vars = { ENV = "production" }

    • Return accounts and account IDs when running wrangler whoami - ashleygwilliams, issue/630 pull/983

      We've made big improvements to wrangler whoami, and now return a list of Cloudflare accounts and account IDs for your authenticated user. If you are unauthenticated, or something is wrong with your API key or token, we'll also return an error with this command to help you understand how to fix your authentication issues!

      Preview

    • Configure sourcemap file - xtuc, issue/681 pull/1063

      webpack (by default) emits a sourcemap that maps to a main.js file, which doesn't match the Workers runtime's configured filename, worker.js. This causes exception reporting tools to be unable to process a Workers sourcemap file – we've updated our Webpack config to output the file worker.js and have fixed this issue.

    • Upload "draft" worker if secret is created before initial worker script has been uploaded - gabbifish, issue/913 pull/1087

      If your script hasn't yet been deployed to the Workers platform, creating and deleting secrets will also create a "draft" Worker – allowing you to still manage secret bindings before you deploy the first version of your script.

  • Maintenance

    • Correctly tar release binaries - EverlastingBugstopper, issue/1055 pull/1062

      This PR updates the way that release binaries are generated during Wrangler's release workflow.

    • Change NPM binary permissions - xtuc, pull/1058

      This PR removes an unnecessary executable permission from npm/binary.js.

    • Improvements to GitHub Actions build process - EverlastingBugstopper, pull/1037

      This PR adds a number of improvements to wrangler's GitHub Actions workflows, including caching, release management, and more granular trigger conditions.

    • Add GitHub Actions badge to README - EverlastingBugstopper, pull/1030

      This PR adds a GitHub Actions badge to our README, indicating whether the repo's builds are currently passing:

      GitHub Actions - Test Status

    • Test Rust with GitHub Actions - EverlastingBugstopper, pull/1028

      This PR adds a GitHub Actions workflow for running wrangler's test suite on a number of platforms and Rust versions.

    • Add release checklist - EverlastingBugstopper, pull/1021

      This PR adds a release checklist, documenting the steps that we use to release new versions of Wrangler. That checklist includes writing this CHANGELOG - very meta!!!

    • Update dependencies - EverlastingBugstopper, pull/1000

      This PR updates some project dependencies as a result of running cargo update.

    • Run CI on pull requests, not pushes - EverlastingBugstopper, pull/1090

      This PR changes the GitHub Actions workflow "event trigger" to fire on pull_request, not push. This will allow wrangler's GitHub Actions workflows to run on PRs sent from forks!

    • Zip .tar files in CI - EverlastingBugstopper, pull/1069 pull/1080

      These PRs fix some issues in the GitHub Actions release workflow that were causing release artifacts to be incorrectly generated.

    • Fixes clippy warnings - EverlastingBugstopper, pull/1071

      This PR fixes some linting issues surfaced by clippy throughout the project.

    • Extract upload and deploy to lib modules - ashleymichal, pull/1075

      This PR refactors some of the underlying code used inside of wrangler publish, to create two higher-level upload and deploy modules. This work has already been used to support "draft workers" in #1087, and to reduce duplication of code between wrangler preview, wrangler dev, and wrangler publish.

💬 1.7.0

  • Features

    • Do not factor in .gitignore into workers sites upload directory traversal - gabbifish, issue/958 pull/981

      This change ensures that the wrangler include/exclude logic for Workers Sites bucket directory traversal does NOT take into account .gitignore, since best practice for static site generators is to put your build directory into your .gitignore.

    • Update cloudflare-rs, reqwest, http, uuid - ashleymichal, issue/301 pull/1009

      These dependency updates may look like routine maintenance, but this reqwest version actually makes support for corporate proxies possible!

    • Add progress bar during Site upload - gabbifish, issue/906 pull/956

      Larger static asset uploads in Wrangler now show a progress bar based on the bulk uploads being made.

    • Allow custom webpack config for Workers Sites projects - ashleymichal, issue/905 pull/957

      Previously we blocked users from declaring webpack_config in their wrangler.toml, as it can be relatively confusing due to the nested nature of the workers-site directory. We removed that block, and added a friendly help message when webpack build fails and the user has a custom webpack_config declared.

    • Reformat config api-key output - bradyjoslin, issue/889 pull/910

      We care a lot about our error output. Now the output from wrangler config is consistent between calls with and without the --api-key flag.

    • Improve error message for wrangler init --site when wrangler.toml already exists - ashleygwilliams, issue/648 pull/931

      wrangler init generally expects that you don't already have a wrangler.toml present; however it is common that users want to add static site functionality to their existing wrangler project and will try using wrangler init to do so. Rather than simply complaining that the toml already exists, now we add the workers-site directory to the project, and print out the suggested configuration to add to wrangler.toml. Much nicer!

    • Add a helpful error message on authentication error - EverlastingBugstopper, issue/492 pull/932

      Previously, when wrangler publish ran into authentication errors, the API result would just print to the screen. Now, it prints a helpful hint to users to re-run wrangler config to fix the error.

    • Provide helpful error when user accidentally puts kv-namespace config under [site] - gabbifish, issue/798 pull/937

      TOML formatting can be tricky, specifically tables, so it is common for users unfamiliar with the format to accidentally nest attributes inside tables without intending it. In this case, if a user adds a kv-namespaces entry to the bottom of a toml with [site] configuration already declared, it is parsed as a part of the [site] table, rather than as a top-level key. The error output from this is not super helpful, as it just says "unknown field kv-namespaces" which isn't precisely correct.

      This PR detects when this error occurs and provides a help suggestion to put kv-namespaces ABOVE the [site] table entry to fix the problem.

  • Fixes

    • Don't install wasm-pack for webpack type projects - EverlastingBugstopper, issue/745 pull/849

      You may have noticed that Wrangler installs wasm-pack for your webpack projects, which may seem strange since it's the tool we use to build Rust projects. The reason for this is because you can also build Rust using wasm-pack and webpack in tandem if you use the wasm-pack-plugin. This plugin recently added support for handling the installation of wasm-pack which means Wrangler no longer needs to handle those installs.

  • Maintenance

    • Make Azure use latest rustc - EverlastingBugstopper, issue/887 pull/893

      Updates our CI to update the rust toolchain to the latest stable version after installation.

    • Fix nightly builds - EverlastingBugstopper, pull/895, pull/898

      Now we confirm Wrangler builds against nightly Rust releases!

    • Fix compiler warnings on windows - uma0317, issue/800 pull/919

      We build Wrangler for Mac OSX, Linux, and Windows, and each of these environments has slightly different needs at compile time. In this case, community contributor uma0317 added configuration that eliminated unused imports for Windows at compile time.

    • Remove deprecated kv-namespace config check - ashleymichal, pull/929

      Back in 1.1.0, we introduced more robust support for adding KV namespaces to your project. It was a breaking change for users who were still using our first pass at configuration for this in their toml, so we added a friendly error message telling them how to update their wrangler.toml. At this point, all of our users have safely transitioned onto the new syntax, and so we removed the warning; any lingering use of the old syntax will be met with a parse error instead.

    • Use binary-install for npm - EverlastingBugstopper, pull/862

      This extracts a lot of the logic in Wrangler's installer to an external package, binary-install, which we will also use for installing wasm-pack on webpack project builds. Switching to this package also has the added benefit of cleaning up the downloaded binary on npm uninstall -g @cloudflare/wrangler.

🎰 1.6.0

  • Features

    • BREAKING CHANGE: Require the webpack_config field in wrangler.toml to build with a custom configuration - EverlastingBugstopper, issue/296 pull/847

      Wrangler will no longer use a webpack.config.js at the root of your project to build your worker. If you would like to continue using a custom build configuration, you will need to specify the webpack_config field in your wrangler.toml like so:

      name = "my-worker"
workers_dev = true
account_id = "01234567890987654321234567890"
webpack_config = "webpack.config.js"

    • API Token Support - gabbifish/ashleymichal, issue/354 pull/471/pull/879

      Wrangler can now be configured with API Tokens!

      Don't worry, current configurations with an email address and a Global API Key will continue to work, but we highly recommend that you switch to API Tokens as they are a much more secure authentication method.

      If you want to use API tokens, create an API token from the "Edit Cloudflare Workers" API token template here, and copy/paste it in the wrangler config prompt. Alternatively, you can set the CF_API_TOKEN environment variable.

    • Add the ability to preview without opening the browser - EverlastingBugstopper, issue/256 pull/816

      wrangler preview can now be called with a --headless flag that will not open the browser.

    • Check for valid credentials when running wrangler config - gabbifish, issue/439 pull/842

      Before this version of Wrangler, wrangler config would allow any input string to be passed for your user details. Now, Wrangler validates that the credentials will work with Cloudflare's API.

    • Add a warning when publishing a Workers Site to a route without a trailing asterisk - EverlastingBugstopper, issue/814 pull/839

      When publishing a Workers Site to your own domain, it's important that the Worker code runs on every path on your domain. This isn't particularly clear, so now when attempting to publish a Workers Site to a route without a trailing asterisk, Wrangler will print a warning message.

    • Better error message for publishing to a duplicate route - pradovic, issue/519 pull/813

      When publishing to a route that is associated with another worker, Wrangler now prints a more actionable error message.

    • Better error message when webpack fails - ashleymichal, issue/428 pull/837

      Wrangler now recommends running npm install as a possible remedy for failed webpack builds.

  • Fixes

    • Properly handle errors when running Wrangler as a global npm package - jaredmcdonald, issue/848 pull/857

    • Clean up temporary build files - EverlastingBugstopper, pull/853

      When building a script, Wrangler creates a temporary file. Old versions of Wrangler were quite messy about it, but now it cleans up after itself.

    • Fix the help text for wrangler generate - EverlastingBugstopper, pull/830

      The default value for a template is now a complete and valid URL instead of a sample project name.

    • Remove --version on subcommands - EverlastingBugstopper, issue/791 pull/829

      Each subcommand in Wrangler used to take a --version argument which would print the name of the subcommand. For instance, wrangler publish --version would print wrangler-publish. This wasn't super helpful, so we've removed that functionality.

    • Fix a broken link in the README - victoriabernard92, pull/838

  • Maintenance

  • Documentation

👻 1.5.0

  • Features

    • Deprecate wrangler publish --release - EverlastingBugstopper, issue/538 pull/751

      wrangler publish --release is now simply an alias of wrangler publish. This is related to the introduction of environments made in 1.3.1 and is intended to reduce confusion surrounding deploy targets. Previously, wrangler publish --release would deploy to a route on your own domain, and wrangler publish would deploy to your workers.dev subdomain. This was a confusing API, and we now require individual environments to have either workers_dev = true or both a route and zone_id in each section of wrangler.toml. This makes it very clear where your Workers code is being deployed. If you're not using wrangler publish --release but you added workers_dev = false to the top level of your wrangler.toml because Wrangler warned you to - you can now safely remove it! If you are using wrangler publish --release, know that it is functionally the same as wrangler publish. If you want to deploy to workers.dev and also a route on your own domain, you will need to set up multiple environments.

    • Deprecate private field in wrangler.toml - stevenfranks, issue/782 pull/782

      In a related note, the private field no longer functions in wrangler.toml. The original intent behind this field was to allow "publishing" without "activating". Unfortunately this led to a lot of undefined behavior if the value was switched from true to false in between a wrangler publish command and a wrangler publish --release command and vice versa. With the removal of wrangler publish --release, we are also removing the private field. If your wrangler.toml files contain a value for private, you can remove it!

    • Include/exclude static assets in a Workers Sites project - gabbifish, issue/716 pull/760

      Your wrangler.toml has two new optional fields: include and exclude. These fields give you more granular control over what files are uploaded to Workers KV. This behavior mirrors Cargo's include/exclude functionality. Further documentation for this feature is available here.

    • A more robust wrangler generate - EverlastingBugstopper, issue/315 pull/759

      wrangler generate is now much smarter about wrangler.toml files. Previously, wrangler generate would simply create the same configuration for every project, and it would ignore any wrangler.toml that was committed to the template. This means much less guesswork when using wrangler generate with existing Workers projects.

    • Add the ability to check if you've already registered a workers.dev subdomain - gusvargas, issue/701 pull/747

      You can now run wrangler subdomain without any arguments to see if you have registered a workers.dev subdomain.

      $ wrangler subdomain
💁  foo.workers.dev

    • Add --verbose flag to wrangler publish and wrangler preview - gabbifish, issue/657 pull/790

      You can now run wrangler publish --verbose and wrangler preview --verbose on a Workers Sites project to view all of the files that are being uploaded to Workers KV.

      $ wrangler publish --verbose
🌀  Using namespace for Workers Site "__example-workers_sites_assets"
💁  Preparing to upload updated files...
🌀  Preparing ./public/favicon.ico
🌀  Preparing ./public/index.html
🌀  Preparing ./public/404.html
🌀  Preparing ./public/img/404-wrangler-ferris.gif
🌀  Preparing ./public/img/200-wrangler-ferris.gif
✨  Success
✨  Built successfully, built project size is 11 KiB.
✨  Successfully published your script to https://test.example.workers.dev

    • Disallow node_modules as a bucket for Workers Sites - gabbifish, issue/723 pull/792

      node_modules is no longer allowed to be a bucket for Workers Sites. It is notoriously very large and if it were specified as a bucket it would probably be a very expensive mistake.

    • Allow installs to utilize Wrangler binaries via a caching proxy instead of GitHub directly - gabbifish, [pull/797]

      To avoid dependency on one external service, GitHub, we enabled a cache proxy (using Workers!) for installations of Wrangler.

    • Provide a better error message when using an unverified email address - ashleygwilliams, issue/320 pull/795

      The Cloudflare API refuses to play nice with unverified email addresses (we don't like spam!), and now when this happens, Wrangler gives an actionable error message.

  • Fixes

    • Fix Rust live preview - gabbifish, issue/618 pull/699

      If you use Wrangler to develop Rust Workers, you may have noticed that live preview (wrangler preview --watch) was not working with your project. Not to worry though, we cracked down on this bug with an (oxidized) iron fist! Wrangler now has cross-platform support for live previewing Rust Workers.

    • Minimize timeout errors for bulk uploads - gabbifish, issue/746 pull/757

      Sometimes Wrangler would make API calls to Workers KV that would timeout if there were too many files. We've increased the amount of time Wrangler will wait around for the API operations to complete.

    • Print readable error message when external commands fail - EverlastingBugstopper, pull/799

      Wrangler depends on a few external applications, and sometimes the calls to them fail! When this happens, Wrangler would tell you the command it tried to run, but it included a bunch of quotes. This change removes those quotes so the command is easily readable and can be copy/pasted.

    • Disallow wrangler generate --site with a template argument - EverlastingBugstopper, issue/776 pull/789

      In Wrangler 1.4.0, we introduced Workers Sites, which included the ability to run wrangler generate --site which would use our site template behind the scenes. Confusingly, you could also pass a template to this command: wrangler generate my-site https://github.com/example/worker-site --site, which would not behave as expected. This specific usage will now correctly output an error.

  • Maintenance

    • Begin refactoring test suite - ashleymichal, pull/787

      We're constantly shipping features in Wrangler, and with more features comes a larger codebase. As a codebase expands, it goes through some growing pains. This release includes some improvements to the internal organization of Wrangler's codebase, and is intended to make our lives and our contributors' lives easier moving forward.

      • Moved all "fixture" helper functions to "utils" module to share between build/preview tests

      • Removed "metadata_wasm.json" from simple_rust fixture

      • Extracted all module declarations in main.rs to lib.rs to allow tests to import with use wrangler::foo

      • Split target/mod.rs into one file per struct

      • Cleaned up KV Namespace mod system

      • Use log::info! instead of info! in main.rs

    • Refactor GlobalUser to be passed as a reference consistently - gabbifish, pull/749

    • Remove internal link from CONTRIBUTING.md - adaptive, pull/784

    • Fix some Clippy warnings - EverlastingBugstopper, pull/793

    • Clean up leftover directories created by tests - ashleymichal, pull/785

    • Refactor subdomain module - EverlastingBugstopper, issue/758 pull/764

    • Fix README markdown misrender - dottorblaster, pull/763

    • Remove duplicate Environments subheader from README - bradyjoslin, pull/766

    • Change Crate author to the Workers Developer Experience team - ashleygwilliams, pull/752

🎂 1.4.0

  • Features

    • Workers Sites - pull/509

      Wrangler 1.4.0 includes supports for Workers Sites, enabling developers to deploy static applications directly to Workers. Workers Sites is perfect for frontend frameworks like React and Vue, as well as static site generators like Hugo and Gatsby.

      Workers Sites is a feature exclusive to Wrangler, and combines a great developer experience with excellent performance. The --site flag has been added to wrangler init and wrangler generate to use Workers Sites with new and existing projects:

      # Add Workers Sites to an existing project
$ wrangler init --site

# Start a new Workers Sites project
$ wrangler generate --site

      If you're ready to get started using Workers Sites, we've written guides for the various routes you might take with your project:

      For more details on how Workers Sites works with Wrangler, check out the documentation. We also have a brand new tutorial to help you learn the Workers Sites workflow, by deploying a React application!

      Workers Sites has been a heroic effort by the entire Workers Developer Experience team, comprising of Wrangler updates, new project templates, and open-source packages. We're super excited about the future that Workers Sites represents, where static sites and serverless functions can work together to build powerful, cutting-edge applications.

      Make sure to try out Workers Sites to build your next app! 🎉🎉🎉

    • Download release from our proxy rather than GitHub - zackbloom, pull/692

    • Add validation for workers names in wrangler init and wrangler generate - gabbifish, issue/470

      There are a number of requirements around what your Workers script is named – previously, Wrangler would opaquely fail and not indicate that your script name was invalid: this PR updates the init and generate commands to validate the potential name before continuing to create a new project.

    • Ensure KV subcommands check for presence of required fields in wrangler.toml - gabbifish, [issue/607] pull/665

      There's a number of commands in wrangler that require a properly configured wrangler.toml file: instead of failing, this PR ensures that these commands now check your configuration file before attempting any action. Hooray for clarity! 😇

    • Show size when a KV upload error occurs - stevenfranks, issue/650 pull/651

      Previously, when uploading a file to Workers KV from Wrangler, the error output didn't indicate the size of the file that was being uploaded. This PR improves the output of that message by showing both the file size and the maximum file size that can be uploaded to Workers KV.

    • Better file support for kv:put - phayes, pull/633

      The kv:put command, introduced in Wrangler 1.3.1, has been improved to support uploading non-UT8 files. In addition, the command now streams files directly to the Cloudflare API when uploading, instead of buffering them in memory during the upload process.

  • Fixes

    Ensure we install and cache the latest version of cargo-generate and wasm-pack if user has an outdated cargo installed version - EverlastingBugstopper, issue/666 pull/726

    Wrangler orchestrates a few other tools under the hood, notably wasm-pack and cargo-generate. We use a library called binary-install to fetch and cache binaries we download. However, to avoid downloading unnecessarily, we first check if the user has a copy locally on their machine that they had cargo install'd. We had a bug where in this logic branch, we didn't check that the local version was the most up-to-date version. This meant that users who had an older installed version may run into errors when wrangler expected to use features of a newer version of that tool. This PR adds the logic to check for the version and will install and cache a newer version for wrangler to use (leaving your local version as is!).

    • Remove link to 000000000000000000.cloudflareworkers.com - EverlastingBugstopper, pull

      Have you ever run wrangler preview in your project and wondered why the URL to preview your application is 000000000000000000.cloudflareworkers.com? The writer of this CHANGELOG finds it confusing, too: this PR removes that line, making it easier to parse the output from wrangler preview.

    • Make install actually fail if the release can't be downloaded - zackbloom, pull/672

      When Wrangler's installation shim attempts to install Wrangler on your machine, there's the possibility that the installation can fail – instead of failing silently, the installation shim now properly throws an error, allowing us to better diagnose installation failures.

  • Maintenance

    • KV command error output improvements - gabbifish, issue/608 pull/613

      The Wrangler team is always on the quest for perfect error messages. In pursuit of that goal, we've improved how errors in the wrangler kv subcommands output to your terminal. 😎

    • Added missing word to whoami response - stevenfranks, pull/695

      Clear writing is good! It's hard to write clearly when words are missing in a sentence. This PR corrects the output of wrangler whoami to add a missing word, making this command easier to read. 🤔

  • Documentation

    • Webpack documentation - EverlastingBugstopper, issue/721 pull/724

      For our default build type, aptly named "webpack", Wrangler uses webpack under the hood to bundle all of your assets. We hadn't documented how we do that, what our default config is, and how you can specify your own custom webpack config if you'd like. We have those docs now, so check them out!

🐛 1.3.1

  • Features

    • Environments - EverlastingBugstopper, issue/385

      Wrangler 1.3.1 includes supports for environments, allowing developers to deploy Workers projects to multiple places. For instance, an application can be deployed to a production URL and a staging URL, without having to juggle multiple configuration files.

      To use environments, you can now pass in [env.$env_name] properties in your wrangler.toml. Here's an example:

      type = "webpack"
name = "my-worker-dev"
account_id = "12345678901234567890"
zone_id = "09876543210987654321"
workers_dev = false

[env.staging]
name = "my-worker-staging"
route = "staging.example.com/*"

[env.production]
name = "my-worker"
route = "example.com/*"

      With multiple environments defined, wrangler build, wrangler preview, and wrangler publish now accept a --env flag to indicate what environment you'd like to use, for instance, wrangler publish --env production.

      To support developers transitioning to environments, we've written documentation for the feature, including further information about deprecations and advanced usage. Check out the documentation here!

    • KV commands - ashleymichal, gabbifish, issue/339

      Wrangler 1.3.1 includes commands for managing and updating Workers KV namespaces, keys, and values directly from the CLI.

      • wrangler kv:namespace

        wrangler kv:namespace allows developers to create, list, and delete KV namespaces, from the CLI. This allows Wrangler users to configure new namespaces without having to navigate into the Cloudflare Web UI to manage namespaces. Once a namespace has been created, Wrangler will even give you the exact configuration to copy into your wrangler.toml to begin using your new namespace in your project. Neat!

        $ wrangler kv:namespace create "MY_KV"
🌀  Creating namespace with title "worker-MY_KV"
✨  Success: WorkersKvNamespace {
  id: "e29b263ab50e42ce9b637fa8370175e8",
  title: "worker-MY_KV",
}
✨  Add the following to your wrangler.toml:
kv-namespaces = [
  { binding = "MY_KV", id = "e29b263ab50e42ce9b637fa8370175e8" }
]

      • wrangler kv:key

        wrangler kv:key gives CLI users access to reading, listing, and updating KV keys inside of a namespace. For instance, given a namespace with the binding KV, you can directly set keys and values from the CLI, including passing expiration data, such as below:

        $ wrangler kv:key put --binding=KV "key" "value" --ttl=10000
✨  Success

      • wrangler kv:bulk

        wrangler kv:bulk can be used to quickly upload or remove a large number of values from Workers KV, by accepting a JSON file containing KV data. Let's define a JSON file, data.json:

        [
  {
    "key": "test_key",
    "value": "test_value",
    "expiration_ttl": 3600
  },
  {
    "key": "test_key2",
    "value": "test_value2",
    "expiration_ttl": 3600
  }
]

        By calling wrangler kv:bulk put --binding=KV data.json, I can quickly create two new keys in Workers KV - test_key and test_key2, with the corresponding values test_value and test_value2:

        $ wrangler kv:bulk put --binding=KV data.json
✨  Success

      The KV subcommands in Wrangler 1.3.1 make it super easy to comfortably query and manage your Workers KV data without ever having to leave the command-line. For more information on the available commands and their usage, see the documentation. 🤯

    • Reduce output from publish command - EverlastingBugstopper, issue/523

      This PR improves the messaging of wrangler publish.

      Before:

      ✨  Built successfully, built project size is 517 bytes.
✨  Successfully published your script.
✨  Success! Your worker was successfully published. You can view it at example.com/*

      After:

      $ wrangler publish
✨  Built successfully, built project size is 523 bytes.
✨  Successfully published your script to example.com/*

    • feat #323: Allow fn & promise from webpack config - third774, issue/323

      This PR fixes Wrangler's handling of webpack.config.js files to support functions and promises, as per the Webpack documentation.

    • Use webworker target - xtuc, issue/477

      This PR updates how Wrangler builds JavaScript projects with Webpack to conform to the webworker build target. This ensures that projects built with Wrangler are less likely to generate code that isn't supported by the Workers runtime.

  • Fixes

    • Have live reload preview watch over entire rust worker directory - gabbifish, pull/535

      This change updates the live reload functionality to watch over the entire Rust worker directory, and does not look for package.json files that do not typically exist for Rust and WASM projects.

    • Fix javascript live preview for linux - gabbifish, issue/517

      This PR fixes an issue with Wrangler's live preview (wrangler preview --watch) functionality on Linux. In addition, it simplifies the console output for a live preview instance, which could get pretty noisy during development!

    • Different emojis for different commands - EverlastingBugstopper, pull/605

      KV subcommands would return the same emoji value in --help output. This PR updates the command-line output to use different emoji, making the output easier to read!

  • Maintenance

    • Add keywords for npm SEO - EverlastingBugstopper, pull/583

      This PR improves the discoverability for wrangler on npm by adding keywords to the installer's package.json.

    • Clean up emoji - xortive, pull/455

      This PR removes some extraneous unicode that keeps our emoji from displaying correctly in certain terminal emulators, especially for multi-codepoint emoji.

  • Documentation

    • Add documentation for init to the README - EverlastingBugstopper, pull/585

      This PR adds documentation in our README for wrangler init, which allows you to begin using an existing project with Wrangler.

    • Remove link to docs for installation because they link back to wrangler README - EverlastingBugstopper, pull/494

    • Minor formatting fix in README.md - kentonv, pull/515

      This PR fixes a small syntax issue in the Wrangler README, causing Markdown lists to render incorrectly.

    • Fix link to Cloudflare Workers preview service - dentarg, pull/472

      This PR fixes a link to the Cloudflare Workers preview service in the Wrangler README.

💆🏻‍♂️ 1.2.0

  • Features

    • Implement live previewing for wrangler - xortive, pull/451

      The wrangler preview command now supports live previewing! As you develop your project, you can start up a link between your local codebase and the preview service by running wrangler preview --watch. Any updates to your project will be passed to the preview service, allowing you to instantly see any changes to your project. This is a massive improvement to the development process for building applications with Workers, and we're super excited to ship it!

      A huge shout-out to @xortive, who almost single-handedly built this feature during his summer internship at Cloudflare, over the last few months. Amazing work! 😍

    • Authenticate calls to preview service when possible - ashleymichal, issue/423 pull/429

      This PR allows developers to use the preview service with account and user-specific functionality, such as KV support inside of the preview service. Previously, attempting to preview a Workers project with KV bindings would cause an error - this PR fixes that, by allowing wrangler preview to make authenticated calls to the preview service.

  • Documentation

  • Fixes

    • Fix ignoring exit code from spawnSync - run-wrangler.js - defjosiah, issue/433 pull/432

      This PR fixes an issue where an NPM-installed wrangler would always return an exit code of 0, even when wrangler had errored. This improves wrangler's ability to be scripted, e.g. in CI projects.

  • Maintenance

    • Test maintenance - EverlastingBugstopper, pull/563

      This PR cleans up some incorrectly named tests and adds fixtures to support testing new functionality in 1.3.1, such as environments. ✨

    • Guard test against potential races - xtuc, pull/567

      This PR fixes a race condition during wrangler build that can occur when two builds are taking place at the same time. To fix it, a file lock has been implemented, which Wrangler will wait on until the previous build completes.

    • Deny clippy warnings in CI, run rustfmt in --check mode - xortive, issue/426

      This PR updates some of the CI steps for testing Wrangler, to make sure that rust-clippy warnings cause CI to fail. This helps Wrangler code stay well-written and consistent - hooray!

    • Nits: fix clippy warnings - xortive, pull/427

    • Add repository link to Cargo.toml - 56quarters, pull/425

🏄‍♀️ 1.1.1

  • Features

    • Install current version, not latest - ashleygwilliams, issue/418

      Previously the NPM installer for wrangler would always pull the most recent release from GitHub releases, and the installer did not increase version numbers when Wrangler did. Many users found this confusing. Now the installer will increment versions along with Wrangler releases, and point at specific versions rather than the most recent one at the time of installation.

    • Improve JSON errors debuggability - xtuc, pull/394

      This PR improves JSON error output in wrangler. Specifically:

      • If a package.json file fails to decode, wrangler now emits a clearer error:

        $ wrangler build
⬇️  Installing wranglerjs...
⬇️  Installing wasm-pack...
thread 'main' panicked at 'could not parse "./package.json": Error("expected `,` or `}`", line: 4, column: 3)', src/libcore/result.rs:999:5

      • If the wranglerjs backend returns invalid JSON, it now preserves the output file for further investigation. Note that the console doesn't print the output file location by default, and you will need to pass RUST_LOG=info while running wrangler build, and search for the --output-file=FILE argument passed to wranglerjs:

        $ RUST_LOG=info wrangler build
⬇️ Installing wasm-pack...
[2019-08-09T19:28:48Z INFO  wrangler::commands::build::wranglerjs] Running "/Users/kristian/.nvm/versions/node/v12.1.0/bin/node" "/Users/kristian/src/workers/wrangler/wranglerjs" "--output-file=/var/folders/5x/yzqyqst11n518yl8xl7yv1f80000gp/T/.wranglerjs_output5eREv" # ...

      • If the preview service returns invalid JSON, it now emits a clearer error, and the full output can be seen by using RUST_LOG=info.

        Previously:

        $ wrangler preview
⬇️ Installing wasm-pack...
⬇️ Installing wranglerjs...
✨   Built successfully.
Error: Error("expected value", line: 2, column: 1)

        Now:

        $ wrangler preview
⬇️ Installing wranglerjs...
⬇️ Installing wasm-pack...
✨   Built successfully, built project size is 1 MiB. ⚠️  Your built project has grown past the 1MiB size limit and may fail to deploy. ⚠️  ✨
Error: https://cloudflareworkers.com/script: Server Error: 502 Bad Gateway

  • Fixes

    • Fix wrangler config for systems with non-unix EOL - xortive, issue/389

      wrangler config was not properly truncating whitespace from the end of user input, resulting in a panic when trying to use wrangler publish, because wrangler would try to create an HTTP header with invalid characters. Now, wrangler will properly truncate extra whitespace (including \r) from the end of the input into wrangler config.

  • Maintenance

    • Migrate straggler emojis to terminal::emoji - EverlastingBugstopper, pull/382

      This PR updates the last remaining instances where wrangler was using hard-coded emojis for messages, rather than using terminal::emoji. In addition, there are two instances where this PR changes the usage of the ⛔ emoji to ⚠️.

    • Move test fixtures to their own directory - EverlastingBugstopper, pull/383

      This PR aggregates fixtures used in integration tests into a fixtures directory to make it easier to find/use them.

    • Update issue templates to fit GitHub's data model - EverlastingBugstopper, pull/387

      Our previous issue templates were not picked up by GitHub's user interface. This PR updates the templates to fit the accepted data model, and adds some style tweaks to make the templates easier to use.

    • Move Emoji formatting/messaging into new functions - ashleymichal, pull/391

      This PR makes improvements to the internal messaging logic of Wrangler, allowing us to be more flexible in how we display information to users.

  • Documentation

    • Update README to include config, env var changes - ashleymichal, pull/379

      In 1.1.0 we changed the config command to be interactive. This PR updates the README to reflect that change.

    • Add section to readme about Vendored OpenSSL - xortive, pull/407

      Wrangler has some external OpenSSL dependencies when installing on Linux -- this PR documents those dependencies, and how to install Wrangler using a vendored OpenSSL feature flag:

      cargo install wrangler --features vendored-openssl

🔑 1.1.0

Wrangler 1.1.0 includes a number of improvements to documentation and project stability, including:

  • Security

    • Change global config perm - xtuc, pull/286

      This PR improves the security of Wrangler's global config file, by restricting access to Wrangler's global user configuration file (~/.wrangler/config/default.toml) to the user who ran the wrangler config command. For existing Wrangler users, please run wrangler config again on your machine! This will fix the permissions of your global config file to be scoped to your user account.

    • Use stdin instead of arguments for wrangler config - xtuc, pull/329

      We've made the wrangler config command interactive – the previous version of the command, wrangler config $email $apiKey, would be captured by your terminal's history, often exposing that information in a ~/.bash_history or a similar file. The new version of wrangler config will prompt you for your email and api_key via stdin.

      In addition, this PR also adds support for a WRANGLER_HOME environment variable, which will be the location for Wrangler's "home" directory, if you need to customize where Wrangler saves its configuration information.

  • Features

    • Support KV Namespace Configuration - ashleymichal, pull/334, add check + error message for pre 1.1.0 kv namespace format - xortive, pull/369

      Wrangler now supports using Workers KV namespaces in your project! To start using KV with your projects, create a namespace in the Cloudflare Dashboard, and the namespace information to your wrangler.toml configuration file. The kv-namespaces key requires setting a binding (the representation of your namespace in your code) and id, the namespace ID:

      # wrangler.toml

[[kv-namespaces]]
binding = "TODOS"
id = "0f2ac74b498b48028cb68387c421e279"

      If you were previously using the undocumented kv-namespaces support in your project config, you'll need to make a few changes to your project to use Wrangler 1.1.0! KV namespaces now need to be created manually, in the Cloudflare Dashboard, to be able to use them in your Wrangler configuration - previous versions of Wrangler created the namespace for you, but the process is now manual, to allow developers to be more explicit about how their KV namespaces are created.

      For users of the previously undocumented kv-namespaces functionality in Wrangler, we've provided a warning and upgrade path, to help you upgrade your KV configuration in your project to the correct format:

      ⚠️  As of 1.1.0 the kv-namespaces format has been stabilized ⚠️
💁‍  Please add a section like this in your `wrangler.toml` for each KV Namespace you wish to bind: 💁‍

[[kv-namespaces]]
binding = "BINDING_NAME"
id = "0f2ac74b498b48028cb68387c421e279"

# binding is the variable name you wish to bind the namespace to in your script.
# id is the namespace_id assigned to your kv namespace upon creation. e.g. (per namespace)

Error: ⚠️  Your project config has an error ⚠️

      This is the initial part of a lot of incredible work being done on supporting Workers KV in Wrangler. If you're interested in what's up next, check out our next milestone.

    • Configure Workers KV in your wrangler.toml - ashleymichal, pull/333

      If you've tried to use Workers KV in Wrangler, you've probably had a bad time! This PR, along with #334, build support for handling and correctly uploading KV namespace information with your Wrangler project:

      [[kv-namespaces]]
binding = "NAMESPACE"
id = "0f2ac74b498b48028cb68387c421e279"

    • Use ENV variables to configure Wrangler - AaronO, pull/225

      Previously, Wrangler required a global configuration file to be able to run. As many users may use Wrangler in situations where they don't have an interactive terminal, meaning they can't instantiate a config file using wrangler config, this PR allows Wrangler to run even if the config file doesn't exist. This change means that users can also configure Wrangler exclusively with environment variables, using $CF_API_KEY for your Cloudflare API key, and $CF_EMAIL for your Cloudflare account email.

    • Adds more descriptive subdomain errors - EverlastingBugstopper, issue/207

      It's super easy to grab a workers.dev subdomain using the subdomain command in wrangler – so easy, in fact, that many people were trying to use it without even having a Cloudflare account! wrangler now warns users when they attempt to add a subdomain without configuring their account_id in wrangler.toml, as well as when you've already registered a subdomain, or if the subdomain you're trying to register has already been claimed.

    • Allow custom webpack configuration in wrangler.toml - EverlastingBugstopper, issue/246

      If you'd like to bring your own Webpack config to your Workers project, you can now specify a webpack_config key in wrangler.toml:

      webpack_config: webpack.prod.js

    • Add issue templates for bug reports and feature requests - gabbifish, issue/250

      To make it easier for us to diagnose problems and support user feedback, we've added issue templates to make it easier for users to submit bug reports and feature requests.

    • Display commands in their defined order - Electroid, pull/236

      We've re-arranged the order of the commands when you run wrangler without any subcommands, so that commonly-used commands are easier to find!

    • Show project size on build - xtuc, pull/205

      Once the build is finished, wrangler now prints the compressed size of the script, and, if available, the Wasm binary size:

      $ wrangler publish
⬇️ Installing wranglerjs...
⬇️ Installing wasm-pack...
✨   Built successfully, built project size is 517 bytes. ✨

    • Add HTTP prefix to publish command output. - elithrar, pull/198

      Prefix "https://" in front of the "script available" output to allow shells to automatically detect it as a link. Many shells will allow you to click directly on the URL from inside the terminal (such as iTerm's "CMD-Click"), making it much easier to navigate to your subdomains, or any published Workers applications.

    • Build: add message and emoji - xtuc, pull/193

      The wrangler team really loves emoji, so we made sure to send a little bit of ✨ cheer ✨ your way, via a new message and emoji, whenever you use the build subcommand. 💌

  • 🤕 Fixes

    • Remove OpenSSL installation step from README - xortive, issue/355

      Due to an OpenSSL dependency in cargo-generate, Wrangler required developers to install OpenSSL as part of the setup process for using Wrangler. Version 0.3.1 of cargo-generate has removed this requirement, and you no longer need to install OpenSSL manually to use Wrangler. Hooray!

    • Fix issue previewing a project with KV namespaces - ashleymichal, pull/353

      This PR fixes a critical bug where using wrangler preview on a project with Workers KV namespaces causes the command to throw an error.

    • Enforce one Webpack entry in configuration - xtuc, pull/245

      wrangler now returns an error during the build process if you use a webpack configuration with more than one export – wrangler needs to have a single known export from webpack to know what to build!

    • Update default template for Rust project type - EverlastingBugstopper, pull/309

      Previously, when passing --type rust to wrangler generate, the only indication that it worked was that the type in wrangler.toml was rust. There were no Rust files in the default template, for a Rust-type project. Now, when running wrangler generate --type rust, wrangler will use rustwasm-worker-template when generating a new project.

    • Stop cleaning webpack build artifacts - EverlastingBugstopper, pull/307

      The configuration for Webpack projects in Wrangler was over-eager to clean build artifacts, which in the worst-case, caused Wrangler to remove source code for developers' projects - oops! This fix relaxes the Webpack cleaning process, to ensure that building a Wrangler project is a bit safer.

    • Correct binding format - xtuc, pull/260

      Previously, wrangler was incorrectly sending up a binding object to the Cloudflare API, whenever we attempted to update a script's bindings. This fix renames it to bindings, and uses an array, as per the Cloudflare API requirements.

    • Correctly pass Wasm module - xtuc, pull/261

      To ensure that a wasm module is successfully passed between wranglerjs and wrangler, the wasm module is now encoded and decoded from base64, avoiding any potential loss of data.

    • Check for account_id and zone_id before publishing - xtuc, issue/170

      The publish subcommand in wrangler now ensures that you have an account_id and zone_id configured in your wrangler.toml file before attempting to publish, instead of failing during the publishing process.

    • Fix Rust ref issue with wranglerjs builds - xtuc, pull/227

      When wranglerjs built a project, it incorrectly referred to the output of that build process without using a Rust reference - this PR fixes that issue and allows wranglerjs to correctly take your bundle, and your project's metadata, and put it all together in a nice little package to send up to the Cloudflare API. Hooray, working projects!

  • 📖 Documentation

    • feat(docs): add CONTRIBUTING.md - ashleygwilliams, pull/268

      We've created a shiny new contribution guide to help contributors understand how the project works, and how the team triages incoming issues and feature requests. This should make it easier for us to work through your feedback on Wrangler, as well as give you some insight into how we work. Woo-hoo! 🎉

    • Update README to include KV config info - ashleymichal, pull/319

      You can now create Workers KV namespaces from inside of your wrangler.toml configuration file - this has been documented in the README.

    • Clarified intro link in README - tomByrer, pull/257

    • Make it more clear that you can install Wrangler though npm - zackbloom, pull/241

    • Document (lightly) the Wrangler 1.0.0 release - signalnerve, pull/204

    • Add README for Wrangler.js package - ashleygwilliams, pull/196

  • 🔧 Maintenance

    • Better error printing - xortive, pull/327

      We've updated how we log errors in Wrangler's output to make it a little easier to read. If you're a Rust developer interested in how we did this, check out the pull request!

    • Always use multipart upload - ashleymichal, issue/280

      We've updated how Wrangler uploads projects to the Workers runtime - as Wrangler supports the entire Workers ecosystem (including tools like Workers KV, the publishing process should use multipart forms to allow uploading all of the data that represents a Workers project in a single step.

    • unify upload form building - ashleymichal, pull/329

      This PR brings consistency to the way that metadata is handled during Wrangler's preview and build commands: previously, a metadata.json file handled the bindings for your Wrangler project, but because it was handled differently for various project types (such as webpack or rust), it led to inconsistent behavior during the upload process. This PR removes usage of metadata.json in favor of building metadata for each project type in-memory, which will then be uploaded directly to Workers platform. This work is foundational for improved Workers KV support in Wrangler 🙌

    • wrangler preview integration tests - EverlastingBugstopper, pull/363

      Wrangler now includes integration tests for wrangler preview, testing every project type that Wrangler supports with our preview service.

    • Add user agent - xtuc, issue/234

      For every outgoing request, wrangler includes a User-Agent header to clearly indicate to servers and APIs that a wrangler client is making a request: wrangler/dev in debug mode and wrangler/$version in release mode.

    • Terminal messaging abstraction - ashleymichal, issue/219

      We've made improvements to Wrangler's terminal output functionality, with support for various log levels and implementations in Wrangler's API for easily using the log levels in future development.

      The new terminal output functionality can be used by importing the terminal::message crate:

      use crate::terminal::message;

message::info("Building project") // "💁‍ Building project"
message::success("Your project has been deployed!") // "✨ Your project has been deployed!"

// Other available functions:
// message::warn, message::user_error, message::working, message::preview

    • Remove pre-push hooks - EverlastingBugstopper, pull/308

      Previous versions of Wrangler included pre-push hooks to ensure that code was linted before being pushed up to Git. This hook made it difficult to manage in-progress work, so the hooks have been removed.

    • Use serde for metadata - xtuc, pull/285

      This change adds proper construction of the worker metadata, previously, it was an error-prone string.

    • Refactor: Conditional per command in main - ashleymichal, pull/279

      The src/main.rs file in Wrangler has been rewritten so that the layout of the file is easier to read.

    • Add an authenticated HTTP client - Electroid, issue/238

      All HTTP requests to the Cloudflare API are now made with an authenticated HTTP client.

    • Move metadata generation at publish-time - xtuc, pull/237

    • Pin webpack version - xtuc, pull/228

      Adds a better control over webpack's version, avoiding possible upstream issues.

    • Remove empty file - xtuc, pull/216

    • test: improve metadata coverage - xtuc, pull/214

    • Reorganize wranglerjs src - xtuc, pull/202 issue/155

    • Minor spelling fix - adaptive, pull/200

👷‍♀️ 1.0.0

Wrangler 1.0.0 has been released! The first major version of Wrangler makes the tool the preferred development and deployment tool for JavaScript and Rust projects to the Cloudflare Workers platform.

This release includes many changes to the developer experience for Wrangler, including:

  • Support for installing wrangler via npm: npm install @cloudflare/wrangler -g.
  • Support for various project types, including javascript and webpack.
    • The default project type for Wrangler is now webpack.
  • Support for creating and binding Workers KV namespaces via your project's configuration file.
  • Enhancements to error messages and validation of your project before building and deploying.
  • Fixes for Windows usage of Wrangler.
  • Fixes for cross-platform console output.

💥 0.1.1

  • 🤕 Fixes

    • Fix publish and preview bug for projects with a - - jaysonsantos, issue/36

      Rust is a sometimes surprisingly opinionated language! When your Cargo.toml specifies a project name with a hyphen(-) in the name, the Rust compiler will implicitly understand this as a _ for all imports and when it creates compiled artifacts it will name them with a _.

      The original implementation of wrangler skipped over this, and as a result would go looking for a wasm file with a - when it should have been looking for a _. This resulted in a bit of a gross error message that stated that a file was not found.

      We've fixed this now- so go ahead and name your packages with -s!

  • 📖 Documentation

    • Install instructions with OpenSSL troubleshooting notes - AustinCorridor, issue/35

      Because of wrangler's use of cargo-generate we use OpenSSL. Classically, this is a tricky dependency. Some users may run into issue with it! We've documented the steps to fix it on MacOS- if you run into this on other platforms, we'd love a PR!

    • Typo and casing fixes - neynah, pull/42

      First releases almost always have some typos. Now they're fixed!

🌌 0.1.0