"The following package identifiers were not found in your indices"

[sjakobi]

~ $ stack --resolver nightly build tasty-stats
Selected resolver: nightly-2017-03-01
Selected mirror https://s3.amazonaws.com/hackage.fpcomplete.com/                                 
Downloading timestamp                                                                            
No updates to your package list were found                                                       
Updating package index Hackage (mirrored at https://s3.amazonaws.com/hackage.fpcomplete.com/) ...The following package identifiers were not found in your indices: tasty-stats-0.2.0.2

This might be related to the recent Hackage maintenance or the S3 downtime. I have notified the Hackage maintainers here.

It would be nice if stack could handle this better or at least make a suggestion like "report a bug".

[mgsloan]

I could indeed reproduce this when you opened it, however, it seems to have resolved itself! Indeed likely due to something like hackage maintenance or s3 downtime.

[sjakobi]

I'm seeing similar exceptions once more:

~/s/hackage-server (master|✔) $ stack init --ignore-subdirs --force
Looking for .cabal or package.yaml files to use to init the project.
Using cabal packages:
- hackage-server.cabal

Selecting the best among 10 snapshots...

Missing some cabal revision files, updating indices
Selected mirror https://s3.amazonaws.com/hackage.fpcomplete.com/                                 
Downloading timestamp                                                                            
No updates to your package list were found                                                       
Did not find .cabal file for safe-exceptions-checked-0.1.0 with SHA of 2d07e15162937261823e35786206c8a218db8ae1 in tarball-based cache
Selected mirror https://s3.amazonaws.com/hackage.fpcomplete.com/                                 
Downloading timestamp                                                                            
No updates to your package list were found                                                       
Missing some cabal revision files, updating indices                                              
Selected mirror https://s3.amazonaws.com/hackage.fpcomplete.com/                                 
Downloading timestamp                                                                            
No updates to your package list were found                                                       
Did not find .cabal file for safe-exceptions-checked-0.1.0 with SHA of 2d07e15162937261823e35786206c8a218db8ae1 in tarball-based cache
Updating package index Hackage (mirrored at https://s3.amazonaws.com/hackage.fpcomplete.com/) ...The following package identifiers were not found in your indices: JuicyPixels-3.2.8.1, alternators-0.1.1.1, bench-1.0.3, cacophony-0.9.2, cmdargs-0.10.15, conduit-combinators-1.1.1, disposable-0.2.0.0, happstack-hsp-7.3.7.2, happstack-server-7.4.6.4, isotope-0.4.0.0, l10n-0.1.0.1, language-c-quote-0.11.7.3, machines-directory-0.2.1.0, matplotlib-0.4.1, microlens-0.4.8.0, microlens-ghc-0.4.8.0, microlens-platform-0.3.8.0, mime-mail-0.4.13.1, mixed-types-num-0.1.0.1, optparse-applicative-0.13.2.0, pipes-attoparsec-0.5.1.5, safe-exceptions-0.1.5.0, safecopy-0.9.3, search-algorithms-0.1.0, stm-extras-0.1.0.2, telegram-api-0.6.1.0, texmath-0.9.3, wai-middleware-auth-0.1.1.1, writer-cps-lens-0.1.0.1, writer-cps-morph-0.1.0.2, yesod-bin-1.5.2.1

(Note the very long last line)

Should I delete some stale cache files or something?

[sjakobi]

Should I delete some stale cache files or something?

I haven't seen these issues anymore since I executed rm -rf ~/.stack/indices but I'm still a bit unhappy about the unhelpful behavior of stack here. I think stack should either make a suggestion on what to do next (delete something?) or attempt to fix the issue itself.

Otherwise the user is just stuck with an unclear exception message and no obvious way to make the command succeed.

[mgsloan]

Strange! Not sure what could lead to that, possibly a bug in hackage-security?

We might consider detecting when it isn't finding anything in the indices, and try deleting them in that case. (not sure that's happening here, but seems like it)

[sjakobi]

Strange! Not sure what could lead to that, possibly a bug in hackage-security?

It seems possible that I still had some funky index/cache data hanging around from haskell/hackage-server#573.

We might consider detecting when it isn't finding anything in the indices, and try deleting them in that case. (not sure that's happening here, but seems like it)

As we don't really understand what's going on we could ask users to make a bug report and append an archive of their ~/.stack/indices – that's probably what I should have done. Afterwards they could manually delete their ~/.stack/indices to fix their immediate problem.

[andrewufrank]

i just saw the same problem (probably after i deleted the lock in .stack/hackage, because i saw repeatedly the same messages "updating package index" with a hint to the lock). when i deleted the full .stack/indices it repopulated the index and now it seems to go ok.

[domenkozar]

Seeing the same issue and removing indices does indeed fix the problem:

$ stack build --nix
There were multiple candidates for the Cabal entry "hlint.hs" (/Users/admin/domenkozar/cardano-sl/.stack-work/downloaded/XTTklt4ROs4T/hlint.hs), picking /Users/admin/domenkozar/cardano-sl/.stack-work/downloaded/XTTklt4ROs4T/tests/hlint.hs
Didn't see serokell-util-0.3 in your package indices.
Updating and trying again.
Selected mirror https://s3.amazonaws.com/hackage.fpcomplete.com/                                 
Downloading timestamp                                                                            
No updates to your package list were found                                                       
Updating package index Hackage (mirrored at https://s3.amazonaws.com/hackage.fpcomplete.com/) ...The following package identifiers were not found in your indices: serokell-util-0.3

[mgsloan]

A couple possibilities here:

1. If the index gets updated, but stack gets killed before deleting the cache, then I think it will be stuck in a state where the cache won't update till the next index update.

2. May be a caching issue in hackage-security

I'm thinking (1) seems plausible. Potential fix: After updating the package index reports no updates, check if the timestamp of the index is newer than the cache. If so, delete the cache. I've opened #3324 with this approach.