Release proposal: v2.0.2 #1679
Release proposal: v2.0.2 #1679
Conversation
|
CI errors currently caused by #1675, not sure if blocker. Likely is that the test is broken, not the patch. |
brendanashworth
added
the
meta
|
5755fc099f should be a noteable change. It's not breaking anything, but should be communicated, imho.
Also, you can strike off the readline known issue. |
|
Blocked until https://jenkins-iojs.nodesource.com/ 's cert is renewed. |
|
Sorry folks, I knew there was something I was missing when I did a recent update of certs! All fixed now. /cc @saghul |
|
Thanks for the heads up! |
|
I'll get this out tomorrow. |
|
Any idea why Jenkins thinks there's such a big delta since the last run on iojs/io.js#master? It's listing 102 new commits! Something didn't get rebased or merged did it? |
|
No? I didn't force-push or anything. @rvagg maybe the updates caused jenkins to loose some history? |
|
no idea! Still lots of child-process failures on Windows, I'm concerned about the quality here. I've asked for a build against v2.0.1 just to see if there's any delta of test failures so we can either blame jenkins, say that it's the same as what we previously released, or genuinely have something to fix: https://jenkins-iojs.nodesource.com/job/iojs+any-pr+multi/677/ |
|
Cool. I was planning on trying to do another one of my "find-windows-bugs" "ci-bisect" tomorrow to try and nail it down. |
|
yrghflbrphlfff! essentially the same failures in v2.0.1 so do we get to assign at least partial blame to Jenkins? @iojs/platform-windows: can any of the Windows collaborators attempt to replicate the failures we're seeing on Windows with child-process? We have little clue why they are just showing up now; although I believe we did have these appearing some time ago as well, I can't recall specifics though. |
|
I don't think we can trust Jenkins at this point, unfortunately. I think some of the failures we're running in to with Edit: I'm not saying that Jenkins is to blame, just that we haven't been able to figure out what goes wrong in the combination of using Jenkins as a builder for our test suite on the buildslaves we've setup. |
|
Again, no windows errors here: https://jenkins-iojs.nodesource.com/view/iojs/job/iojs+any-pr+multi/651/ Windows errors here: (uh, what happened to my original comparison CI? Here's a new one): https://jenkins-iojs.nodesource.com/view/iojs/job/iojs+any-pr+multi/678/ Jenkins appears to be messing with us again. |
|
Still been too busy today. Someone else could do the release if they'd like, otherwise I'll try to do it tomorrow. |
|
I'm out, no time today; @chrisdickinson it's up to you if you want to handle making a changelog entry and pushing release today if you want |
Notable changes
Known issuesSee https://github.com/iojs/io.js/labels/confirmed-bug for complete and current list of known issues.
Going to also merge #1694 and then convert this to a PR + do the things, can someone review the changelog? |
|
For snapshots, may want to mention how this should result in a big performance improvement for vm context creation. (And startup, although I imagine startup is pretty slow right now anyway...) Still find the known issues list ridiculous, but whatevs. |
why? |
|
See #1484. |
|
Updated changelog:
@domenic lgty? |
|
LGTM |
PR-URL: nodejs#1679 Notable Changes: * win,node-gyp: the delay-load hook for windows addons has now been correctly enabled by default, it had wrongly defaulted to off in the release version of 2.0.0 (Bert Belder) nodejs#1433 * os: tmpdir()'s trailing slash stripping has been refined to fix an issue when the temp directory is at '/'. Also considers which slash is used by the operating system. (cjihrig) nodejs#1673 * tls: default ciphers have been updated to use gcm and aes128 (Mike MacCana) nodejs#1660 * build: v8 snapshots have been re-enabled by default as suggested by the v8 team, since prior security issues have been resolved. This should give some perf improvements to both startup and vm context creation. (Trevor Norris) nodejs#1663 * src: fixed preload modules not working when other flags were used before --require (Yosuke Furukawa) nodejs#1694 * dgram: fixed send()'s callback not being asynchronous (Yosuke Furukawa) nodejs#1313 * readline: emitKeys now keeps buffering data until it has enough to parse. This fixes an issue with parsing split escapes. (Alex Kocharin) * cluster: works now properly emit 'disconnect' to cluser.worker (Oleg Elifantiev) nodejs#1386 events: uncaught errors now provide some context (Evan Lucas) nodejs#1654
PR-URL: nodejs#1679
|
@piscisaureus earlier this week you pushed up a |
|
@Fishrock123 the armv6 build failed because of a misconfiguration on the build slave that ended up doing it, I've manually fixed that configuration and also manually uploaded the files to staging so you can promote and sign them now. |
|
Done, thanks. |
PR-URL: nodejs#59 Related: nodejs/node#1679 Signed-off-by: Hans Kristian Flaatten <hans.kristian.flaatten@turistforeningen.no>
PR-URL: nodejs/node#1679 Notable Changes: * win,node-gyp: the delay-load hook for windows addons has now been correctly enabled by default, it had wrongly defaulted to off in the release version of 2.0.0 (Bert Belder) #1433 * os: tmpdir()'s trailing slash stripping has been refined to fix an issue when the temp directory is at '/'. Also considers which slash is used by the operating system. (cjihrig) #1673 * tls: default ciphers have been updated to use gcm and aes128 (Mike MacCana) #1660 * build: v8 snapshots have been re-enabled by default as suggested by the v8 team, since prior security issues have been resolved. This should give some perf improvements to both startup and vm context creation. (Trevor Norris) #1663 * src: fixed preload modules not working when other flags were used before --require (Yosuke Furukawa) #1694 * dgram: fixed send()'s callback not being asynchronous (Yosuke Furukawa) #1313 * readline: emitKeys now keeps buffering data until it has enough to parse. This fixes an issue with parsing split escapes. (Alex Kocharin) * cluster: works now properly emit 'disconnect' to cluser.worker (Oleg Elifantiev) #1386 events: uncaught errors now provide some context (Evan Lucas) #1654
PR-URL: nodejs/node#1679
As suggested by @rvagg at #1673 (comment). Contains fixes. Proposing this within ~48h.
8a0e5295b4] - build: use backslashes for paths on windows (Johan Bergström) #169820c9a52227] - build: move --with-intl to intl optgroup (Johan Bergström) #168036cdc7c8ac] - build: re-enable V8 snapshots (Trevor Norris) #16635883a59b21] - cluster: disconnect event not emitted correctly (Oleg Elifantiev) #13860f850f7ae7] - deps: provide TXT chunk info in c-ares (Fedor Indutny)7e1c0e75ed] - deps: sync with upstream c-ares/c-ares@bba4dc5 (Ben Noordhuis) #167818d457bd34] - dgram: call send callback asynchronously (Yosuke Furukawa) #13138b9a1537ad] - events: provide better error message for unhandled error (Evan Lucas) #165419ffb5cf1c] - lib: fix eslint styles (Yosuke Furukawa) #153976937051f8] - os: refine tmpdir() trailing slash stripping (cjihrig) #1673aed6bce906] - readline: turn emitKeys into a streaming parser (Alex Kocharin) #16010a461e5360] - src: fix preload when used with prior flags (Yosuke Furukawa) #1694931a0d4634] - src: add type check to v8.setFlagsFromString() (Roman Klauke) #165208d08668c9] - src,deps: replace LoadLibrary by LoadLibraryW (Cheng Zhao) #2264e2f999a62] - test: fix infinite loop detection (Yosuke Furukawa) #16815755fc099f] - tls: update default ciphers to use gcm and aes128 (Mike MacCana) #1660966acb9916] - tools: remove closure_linter to eslint on windows (Yosuke Furukawa) #1685c58264e58b] - tools: make eslint work on subdirectories (Roman Reiss) #16860b21ab13b7] - tools: refactormake test-npminto test-npm.sh (Jeremiah Senkpiel) #1662f07b3b600b] - tools: set eslint comma-spacing to 'warn' (Roman Reiss) #1672f9dd34d301] - tools: replace closure-linter with eslint (Yosuke Furukawa) #153964d3210c98] - win,node-gyp: enable delay-load hook by default (Bert Belder) #1667NOTE:
64d3210c98] - (SEMVER-MAJOR) win,node-gyp: enable delay-load hook by default (Bert Belder) #1433Is not technically SEMVER-MAJOR here even though it is labeled as such. This is a bug fix for it NOT being correctly turned on in 2.0.0. The commit however, does not have the correct meta-data and the new issue is actually #1667. This has been updated in the changelog.