Creating new version failed: Application error

[Gawdl3y]

I'm currently receiving this error message when attempting to publish the latest update for my package, language-papyrus:

> apm publish --tag v1.2.0
Publishing language-papyrus@v1.2.0 failed
Creating new version failed: Application error

I have used apm login to authorise multiple times to ensure I pasted the API key in correctly, but no dice.

> apm -v
apm  1.9.2
npm  2.13.3
node 0.10.40
python
git 2.8.1.windows.1
visual studio

[Gawdl3y]

I went ahead and skipped 1.2.0 and went straight to 2.0.0 since it makes a decent amount of sense anyways, and publishing was successful for 2.0.0. It was only 1.2.0 that refused to publish.

[mjansing]

I'm also running into this issue. I updated language-rdf to v0.4.0 and cannot publish this version:

$ apm publish 0.4.0
Preparing and tagging a new version ✓
Pushing v0.4.0 tag ✓
Publishing language-rdf@v0.4.0 ✗
Creating new version failed: Application error

After that apm publish --tag v0.4.0 also runs into this error:

$ apm publish --tag v0.4.0
Publishing language-rdf@v0.4.0 ✗
Creating new version failed: Application error

What can I do? Is there any option to get verbose logging output?

[mjansing]

Any updates on this issue? I'm still unable to publish my updated package.

[mjansing]

I solved it by skipping version v0.4.0 and publishing v0.5.0.

[ioquatix]

This appears to still be broken.

[ioquatix]

^_^ > apm publish minor
Preparing and tagging a new version ✓
Pushing v2.4.0 tag Username for 'https://github.com': ioquatix
Password for 'https://ioquatix@github.com': 
✓
Publishing script-runner@v2.4.0 ✗
Creating new version failed: Application error

[eelstork]

Seeing the same. (yes, did login first). Don't think this is just one error, seems to be a catch-all error message for a variety of conditions. Would be nice to document a way to detail what goes on here.

[eelstork]

Error remains.

However I could verify that the package was updated (though not every publish action did result in an update).

...and, assuming this did work as I did not question what the pkg manager web UI is saying).

Simultaneously package search Web UI returns 5⚛︎0 so, maybe something there? But that's the thing. If the service is down on host side it would be nice for apm to point this out.

I'll go fish for the apm source if anybody would gallantly point me at it thank you aplenty.

[ThatXliner]

I have this problem right now

[d-Pixie]

As do I 🙁

Have tried minor, patch and major as well as specific tag. I also can not unpublish versions properly - I get the same error then.

[ThatXliner]

I haven’t tried degrading apm yet

[d-Pixie]

I also tried to run the commands over the API, but I get the same, unhelpful. error back then. I'm starting to wonder if something with the expected package structure has changed. Mine is an old package that I'm updating for the first time in years - so maybe something is supposed to be different now. It works locally, in dev mode, but I'll inspect another package and see if I can see something structurally different. One would think that should be part of the apm publish preflight though ... I also noted that several things are broken around atom.io. The discuss link in some places doesn't work, the contact form gives 500, some documentation link I tried was broken ... So maybe this isn't as well maintained as I would have thought?

[d-Pixie]

@sadick254 could you - or someone else from the team - shed some light on this please?

[rolfedh]

This issue is a blocker for a doc team that needs to upgrade to a package with a bug fix in it: https://github.com/errata-ai/vale-atom/issues/8

[godlikemouse]

I'm having the same issue. Using the latest apm running apm publish minor gives the error: Creating new version failed: Application error

[miyako]

Hi.

I am not sure, but I think I stumbled across a workaround. it goes like

1. apm publish minor (or whatever)
2. ignore application error
3. apm publish minor (again)
4. apm unpublish {your-plackage@the latest semantic}

in other words, 2 steps forward, 1 step back.

[godlikemouse]

@miyako Thank you for the info, but for some reason this doesn't work for me. I tried pushing to minor updates, received the application error on both, ignored. used apm unpublish language-mads-6502-asm@v0.17.0 (which is my latest) and received the following error:

Unpublishing language-mads-6502-asm@v0.17.0 ✗
Unpublishing failed: Not found

[godlikemouse]

Right now I'm completely locked and my package is unusable by anyone. Because the publish failed, it left it in a state that makes it look like there's an update, but it can't be installed.

[4D-JP]

Unpublishing language-mads-6502-asm@v0.17.0

in this context you need to remove the "v"

the command should be

apm unpublish language-mads-6502-asm@0.17.0

[godlikemouse]

Wonderful, thank youo @4D-JP and @miyako . That worked and for now at least, I have a way to get updates out using this work around. Hopefully, the publish will get fixed in the near future.

[willflet]

I'm also having this issue. Replicated on Mac, Windows, and WSL.

[marcelkohl]

Confirmed here also that @miyako workaround works. (Publish new version, unpublish the old version) Check the latest published version with apm view your-package-name

[xavierfoucrier]

Hello everyone, trying to publish from minor version to major one, no luck with the workaround provided. You confirmed this is related to apm / atom.io?

[idleberg]

Can't say the workaround does anything for me.

[DeeDeeG]

Basically, the Atom packages API server has been somewhat unreliable lately, and it has been intermittently returning 500 internal server error seemingly at random. I think that's the real problem here.

So, the workaround is to try again a couple of times, hope you don't get a 500 internal server error again, and eventually you should get a successful publish. (Then, if needed, clean up/apm unpublish any broken versions that might have been partially uploaded to the packages API server?)

That's what @miyako already said, but I hope this explains why it works and explains what is going on. This is my best understanding of the situation, since there hasn't been an official explanation of the situation given yet, that I've seen.