crashing bug in prerelease version parser #3147

samhatoum · 2014-11-21T17:07:20Z

Yesterday I published a velocity:core with version set to 1.0.0-rc.6-cucumber. Since then, all packages that depend on previous versions of velocity:core are no longer working, the error displayed is shown below.

To replicate try to add mike:mocha to any meteor project. Here's a copy/paste repro:

meteor create semvarIssue
cd semvarIssue
meteor add mike:mocha

You will see the following:

Error: Unexpected character in prerelease identifier: 6
    at validCharToNumber (packages/package-version-parser/package-version-parser.js:102)
    at packages/package-version-parser/package-version-parser.js:108
    at Array.reduce (native)
    at Function._.reduce._.foldl._.inject (packages/underscore/underscore.js:139)
    at prereleaseIdentifierToFraction (packages/package-version-parser/package-version-parser.js:92)
    at Object.PV.versionMagnitude (packages/package-version-parser/package-version-parser.js:82)
    at packages/constraint-solver/constraint-solver.js:424
    at Object.each (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/isopacks/constraint-solver/npm/node_modules/mori/mori.js:320:439)
    at Object.resolverOptions.costFunction (packages/constraint-solver/constraint-solver.js:388)
    at overallCostFunction (packages/constraint-solver/resolver.js:200)
    at packages/constraint-solver/resolver.js:232
    at Array.forEach (native)
    at Function._.each._.forEach (packages/underscore/underscore.js:105)
    at ConstraintSolver.Resolver.resolve (packages/constraint-solver/resolver.js:228)
    at ConstraintSolver.PackagesResolver.resolve (packages/constraint-solver/constraint-solver.js:238)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/catalog.js:327:32
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:338:18
    at _.extend.withValue (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/fiber-helpers.js:112:14)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:331:34
    at _.extend.withValue (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/fiber-helpers.js:112:14)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:329:23
    at _.extend.withValue (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/fiber-helpers.js:112:14)
    at Object.enterJob (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:303:26)
    at _.extend.resolveConstraints (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/catalog.js:320:30)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/commands-packages.js:2219:38
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:250:13
    at _.extend.withValue (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/fiber-helpers.js:112:14)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:243:29
    at _.extend.withValue (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/fiber-helpers.js:112:14)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:241:18
    at _.extend.withValue (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/fiber-helpers.js:112:14)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:232:23
    at _.extend.withValue (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/fiber-helpers.js:112:14)
    at Object.capture (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/buildmessage.js:231:19)
    at main.registerCommand.name [as func] (/Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/commands-packages.js:2212:33)
    at /Users/sam/.meteor/packages/meteor-tool/.1.0.35.gxa3d3++os.osx.x86_64+web.browser+web.cordova/meteor-tool-os.osx.x86_64/tools/main.js:1353:23

The text was updated successfully, but these errors were encountered:

ghost · 2014-11-21T17:11:52Z

Maybe the - character before cucumber is illegal in this part.

samhatoum · 2014-11-21T17:12:45Z

Yep, and it looks like there might have been a fix for this?

#2715

skandragon · 2014-11-21T17:20:44Z

And yesterday, when I first tried Velocity, it worked. Today, it does not. :P Great timing! I was just about to add it to another dozen projects or so.

In any case, I too am getting this error. I thought the referenced fix went into a pre-1.0 meteor, but apparently either it did not fix it, or it is a new issue. It's not clear what was fixed, as no commits reference that issue number.

skandragon · 2014-11-21T17:23:18Z

What if it's called 1.0.0-rc6-cucumber? That is, without the .6?

awatson1978 · 2014-11-21T17:24:07Z

It's the dash character. The dash is reserved for release suffix, and this creates two suffixes.

skandragon · 2014-11-21T17:24:43Z

It seems that anything after the first dash should be the suffix...

awatson1978 · 2014-11-21T17:26:53Z

Depends on how the regex is written.

samhatoum · 2014-11-21T17:27:15Z

I tried two new releases right after that, which are:

Version 1.0.0-rc.6-cucumber.0
Version 1.0.0-temp.0

None of them supercede the previously added two dashes

skandragon · 2014-11-21T17:29:50Z

From the semantic versioning spec 2.0.0:

Identifiers MUST comprise only ASCII alphanumerics and hyphen [0-9A-Za-z-].

It looks like meteor's parser is not following the spec, regardless of how the regexp is written. :)

Plus, with the ability to have build info (with a + marker) it looks like a simple parser using a regexp would be out of the question at this point. It looks like it'll have to become stateful, and follow character by character to determine what it is parsing as it finds the first and following dots, dash, and plus sign.

skandragon · 2014-11-21T17:39:08Z

Without actually doing it yet... I'd probably write this as:

Split at the first + sign found. Discard the part after it.
Split at the first - sign found. If there is anything after it, this is the prerelease identifier.
Process the remainder as major.minor.patch format data.

The prerelease part may need to be further broken up by dots later, if the major.minor.patch comparison is equal, since "alpha" < "alpha.1".

ghost · 2014-11-21T17:43:03Z

I think the numbers 0-9 are missing in https://github.com/meteor/meteor/blob/release/METEOR%401.0/tools/package-version-parser.js#L97?

skandragon · 2014-11-21T17:47:55Z

I don't think so. It looks like numbers are intended to be caught in the "if (typeof part === 'number')" case.

skandragon · 2014-11-21T17:52:14Z

I think the real problem here is that pre-release IDs are probably able to be treated as simple strings for comparison, and instead it's trying to do magic.

Of course, that there are dots inside a pre-release ID that matter, makes this harder. I guess one could compare pure numeric fields as numeric, and any field that has a string as a string, but I have no idea what happens when you try to compare "alpha.0" and "alpha.b"

ghost · 2014-11-21T17:52:16Z

I think the semver package parses the version to: { ..., prerelease: ['rc', '6-cucumber']}.
Therefore the second part with the 6 is no number.

skandragon · 2014-11-21T17:53:52Z

So, to get this package working again for those of us wanting to use it... can Version 1.0.0-rc6-cucumber and rc.6-cucumber.0 be removed from the package system entirely?

ghost · 2014-11-21T17:59:47Z

A temporary fix is to add velocity:core@1.0.0-rc.6 to your .meteor/packages file and your .meteor/versions file.

meteor update will fail though. You can temporarely remove all testing packages from .meteor/packages if you need to execute meteor update.

rissem · 2014-11-21T18:04:31Z

Here's a simple reproduction repo I just made.

skandragon · 2014-11-21T18:21:17Z

Can't each of the elements of a pre-release ID (that is, the parts between the dots) be treated as a string, and directly compared that way? "1" < "100", "10" < "100", etc. This appears to be the intent of the spec, but it's not clear on this. Thus, if you had an array:

["rc", "6-cucumber"] it would compare to ["rc", "6"] and come after it automatically and predictability.

rissem · 2014-11-21T18:27:33Z

I just published velocity:core@0.3.0 (equivalent to rc5) as a work around for this issue.

skandragon · 2014-11-21T19:32:36Z

Success! There is no way to un-publish something once it's published? That seems like a flaw.

samhatoum · 2014-11-21T19:37:29Z

Totally. I'm glad we've found an issue and I'm sure the MDG will fix it. Here's a story I'd love to see implemented:

As a package maintainer
I would like to remove/unpublish my packages
So that I can keep my users happy

:)

glasser · 2014-11-21T19:57:41Z

OK, does that mean there's no current emergency issue here?

We can remove packages right now but it's not the greatest operation (eg it ends up requiring a full catalog resync for all users due to how the sync protocol works). And we'll look into the underlying bug soon too. (I apologize, I'm generally in charge of our GitHub bug triage process and I've been taking a few weeks of focusing on fixing [a few problems](https://github.com/meteor/meteor/milestones/Tool Performance and Stability) instead of incoming bugs!)

ghost · 2014-11-21T20:17:09Z

OK, does that mean there's no current emergency issue here?

Yes. We have solved our issue, that velocity:core was no longer usable, by releasing a new version 0.3.0.

samhatoum mentioned this issue Nov 21, 2014

Unable to install this meteor module (directly or via dependency) meteor-velocity/html-reporter#48

Closed

rissem mentioned this issue Nov 21, 2014

Invalid package RC versions break all other RCs #3150

Closed

rissem mentioned this issue Nov 21, 2014

Unexpected character in prerelease identifier mad-eye/meteor-mocha-web#97

Closed

brishin mentioned this issue Nov 21, 2014

Readd velocity core when semver error is fixed adicu/Courses#245

Open

glasser changed the title ~~velocity:core dependents no longer working due to semvar issue~~ crashing bug in prerelease version parser Dec 4, 2014

glasser mentioned this issue Dec 4, 2014

test-packages fails on travis when repository name doesn't match package name #3096

Closed

glasser closed this as completed in 44563f1 Dec 10, 2014

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

crashing bug in prerelease version parser #3147

crashing bug in prerelease version parser #3147

samhatoum commented Nov 21, 2014

ghost commented Nov 21, 2014

samhatoum commented Nov 21, 2014

skandragon commented Nov 21, 2014

skandragon commented Nov 21, 2014

awatson1978 commented Nov 21, 2014

skandragon commented Nov 21, 2014

awatson1978 commented Nov 21, 2014

samhatoum commented Nov 21, 2014

skandragon commented Nov 21, 2014

skandragon commented Nov 21, 2014

ghost commented Nov 21, 2014

skandragon commented Nov 21, 2014

skandragon commented Nov 21, 2014

ghost commented Nov 21, 2014

skandragon commented Nov 21, 2014

ghost commented Nov 21, 2014

rissem commented Nov 21, 2014

skandragon commented Nov 21, 2014

rissem commented Nov 21, 2014

skandragon commented Nov 21, 2014

samhatoum commented Nov 21, 2014

glasser commented Nov 21, 2014

ghost commented Nov 21, 2014

crashing bug in prerelease version parser #3147

crashing bug in prerelease version parser #3147

Comments

samhatoum commented Nov 21, 2014

ghost commented Nov 21, 2014

samhatoum commented Nov 21, 2014

skandragon commented Nov 21, 2014

skandragon commented Nov 21, 2014

awatson1978 commented Nov 21, 2014

skandragon commented Nov 21, 2014

awatson1978 commented Nov 21, 2014

samhatoum commented Nov 21, 2014

skandragon commented Nov 21, 2014

skandragon commented Nov 21, 2014

ghost commented Nov 21, 2014

skandragon commented Nov 21, 2014

skandragon commented Nov 21, 2014

ghost commented Nov 21, 2014

skandragon commented Nov 21, 2014

ghost commented Nov 21, 2014

rissem commented Nov 21, 2014

skandragon commented Nov 21, 2014

rissem commented Nov 21, 2014

skandragon commented Nov 21, 2014

samhatoum commented Nov 21, 2014

glasser commented Nov 21, 2014

ghost commented Nov 21, 2014