Fix brig doc module generation

[pcapriotti]

Follow-up to #1956, fixing issues reported by @fisx:

• remove obsolete data-files stanza from brig's cabal file;
• generate doc modules on repl invocation, as well as on build.

No CHANGELOG entry.

Checklist

• The PR Title explains the impact of the change.
• The PR description provides context as to why the change should occur and what the code contributes to that effect. This could also be a link to a JIRA ticket or a Github issue, if there is one.

[fisx]

No CHANGELOG entry.

You could manually add the two PRs to the one changelog entry you probably (hopefully?) made about the change you're fixing here?

[fisx]

I still get the same error, but I'm not sure how to reproduce it without a 200 year old emacs haskell-mode and a self-written ghci wrapper script. stack repl works fine.

But... how?! And why did having the data file in the cabal file break things?

[fisx]

don't files that are required for compilation also count as data files?

[pcapriotti]

I don't think so.

[fisx]

but they should make it into the tar ball created by cabal sdist, no? and that's only happening if they are listed here.

$ cabal sdist
remote: Enumerating objects: 60, done.
remote: Counting objects: 100% (60/60), done.
remote: Compressing objects: 100% (25/25), done.
remote: Total 60 (delta 40), reused 55 (delta 35), pack-reused 0
Unpacking objects: 100% (60/60), 14.09 KiB | 721.00 KiB/s, done.
From https://github.com/wireapp/http2
 * [new branch]      aeson-2             -> origin/aeson-2
 * [new branch]      client-status-fixme -> origin/client-status-fixme
 * [new branch]      preface-race        -> origin/preface-race
 * [new branch]      smatting/debugging  -> origin/smatting/debugging
Previous HEAD position was 7c465be Better support for setting header table size to 0
HEAD is now at 1ee1ce4 Make sure connection preface is always sent first
cabal: Error: Could not find module: Brig.Docs.Swagger with any suffix:
["gc","chs","hsc","x","y","ly","cpphs","hs","lhs","hsig","lhsig"]. If the
module is autogenerated it should be added to 'autogen-modules'.

ah, the last line is interesting!

[fisx]

(yes, we don't do sdists, but we might as well pretend! :))

[fisx]

and why did you modify the cabal file instead of package.yaml?

[pcapriotti]

This only ended up by mistake in the cabal file, not in the package.yaml file.

changed my mind

[fisx]

concourse failed for the same reason #1964 failed earlier today:

  test/integration/Test/Federator/InwardSpec.hs:77:9: 
  1) Federator.API.Inward should be able to call cargohold
       uncaught exception: ErrorCall
       Assertions failed:
        1: 403 =/= 500
       
       Response was:
       
       Response {responseStatus = Status {statusCode = 500, statusMessage = "Internal Server Error"}, responseVersion = HTTP/1.1, responseHeaders = [("Transfer-Encoding","chunked"),("Date","Wed, 08 Dec 2021 13:56:38 GMT"),("Server","Warp/3.3.17"),("Content-Type","application/json")], responseBody = Just "{\"code\":500,\"message\":\"Federation is not yet implemented for this endpoint\",\"label\":\"federation-not-implemented\"}", responseCookieJar = CJ {expose = []}, responseClose' = ResponseClose}
       CallStack (from HasCallStack):
         error, called at src/Bilge/Assert.hs:89:5 in bilge-0.22.0-FhngfA5b3MW5Tg8nd5Ja7l:Bilge.Assert
         <!!, called at src/Bilge/Assert.hs:107:19 in bilge-0.22.0-FhngfA5b3MW5Tg8nd5Ja7l:Bilge.Assert
         !!!, called at test/integration/Test/Federator/InwardSpec.hs:77:9 in main:Test.Federator.InwardSpec

is it possible that #1956 or #1973 broke develop?

[pcapriotti]

I still get the same error, but I'm not sure how to reproduce it without a 200 year old emacs haskell-mode and a self-written ghci wrapper script. stack repl works fine.

If you can share a script that breaks, I'm happy to look into it. But both stack repl and cabal repl work for me.

But... how?! And why did having the data file in the cabal file break things?

No, that was unrelated, just some harmless leftover. The fix is to generate the doc files on the repl hook, as well as on the build hook.