Set PATH_SEPARATOR=; when calling ./configure; fix #7494

[Mikolaj]

No tests, because they depend on a particular version of autoconf and I think the current tests of autoconf would already fail with the new autoconf version on Windows CI (but either they are not run on Windows or autoconf on CI is old; they are run on Linux, because setting PATH_SEPARATOR breaks these tests and @jneira kindly reproduced the fail on Windows locally).

Edit: what I missed here and what led to regression in cabal 3.6 is that packages can carry their own generated autoconf files, from versions we have no control of (e.g., which did not run with "PATH_SEPARATOR=;" nor did predict such usage, but now we are using them with "PATH_SEPARATOR=;"). And I bet the story is even more complex.

---

Please include the following checklist in your PR:

• Patches conform to the coding conventions.
• Any changes that could be relevant to users have been recorded in the changelog.
• The documentation has been updated, if necessary.

Please also shortly describe how you tested your change. Bonus points for added tests!

[jneira]

so we should test it locally I suppose, will do asap

[Mikolaj]

@jneira: thank you. Please try with the second commit to make sure my CPP works fine.

[jneira]

nice, I already reproduced the error in my windows 10 just in case.
do we know what other important packages other than network uses configure?

[Mikolaj]

Perhaps old-time? Just guessing. No other mentions in the issue.

[phadej]

Also time and `, if you reinstall them, i.e. don't use the versions bundled with GHC.

https://hackage-search.serokell.io/?q=build-type%3A.*Configure

[fgaz]

Suggested change

	- Set PATH_SEPARATOR=";" for ./configure on Windows; fix for new autoconf
	- Set PATH_SEPARATOR=";" for ./configure on Windows; fix for autoconf >= 2.70

(it can also just go all in the synopsis with no extended description)

[Mikolaj]

Heh, the details didn't fit in 80 lines, despite condensing wrt synposis. Let me go full hog in synopsis, after your encouragement.

[fgaz]

Oh, there's no hard limit on synopsis. It's just converted into changelog bullet points (or sub-points for description) after all

[jneira]

https://hackage-search.serokell.io/?q=build-type%3A.*Configure

😮 i did not know it, pure gold, thanks!

[jneira]

• It works in my windows 10 with network:

atrey@FNTSY MINGW64 /d/dev/ws/haskell/issues/network
$ cabal --version
cabal-install version 3.4.0.0
compiled using version 3.4.0.0 of the Cabal library

atrey@FNTSY MINGW64 /d/dev/ws/haskell/issues/network
$ cabal build
Resolving dependencies...
Build profile: -w ghc-8.10.5 -O1
In order, the following will be built (use -v for more details):
 - network-3.1.2.2 (lib:network) (first run)
Configuring network-3.1.2.2...
configure: WARNING: unrecognized options: --with-compiler
configure: loading site script D:/ghcup/msys64/etc/config.site
configure: error: cannot find required auxiliary files: config.guess config.sub

atrey@FNTSY MINGW64 /d/dev/ws/haskell/issues/network
$ /d/bin/cabal.exe --version
cabal-install version 3.5
compiled using version 3.5.0.0 of the Cabal library

atrey@FNTSY MINGW64 /d/dev/ws/haskell/issues/network
$ /d/bin/cabal.exe build
Resolving dependencies...
Build profile: -w ghc-8.10.5 -O1
In order, the following will be built (use -v for more details):
 - network-3.1.2.2 (lib:network) (first run)
Configuring network-3.1.2.2...
configure: WARNING: unrecognized options: --with-compiler
configure: loading site script D:/ghcup/msys64/etc/config.site
checking build system type... x86_64-w64-mingw32
checking host system type... x86_64-w64-mingw32
checking for gcc... D:\ghcup\ghc\8.10.5\lib\../mingw/bin\gcc.exe
....

I will try another pair of projects

[jneira]

I've tried with process and both versions of cabal works as intended. I think the cause is for process there are no config.guess and config.sub files.

[Mikolaj]

@jneira: thank you so much. Would you and @fgaz review so that I can merge?

[fgaz]

What about the TODO?

[Mikolaj]

@fgaz: this TODO is a style thing, and I still don't quite feel the style of this codebase, so I'd need a guidance. What would be your recommendation? Keep the sole CPP (ATM it doesn't look like it could grow or multiply) or create a value in a Compat module and, if so, in which one?

[jneira]

I did not test it with an older autoconf version, I have to downgrade it manually to do it

[jneira]

I did not test it with an older autoconf version, I have to downgrade it manually to do it

atrey@FNTSY MINGW64 /d/dev/ws/haskell/issues/network
$ autoreconf --version
autoreconf (GNU Autoconf) 2.69
Copyright (C) 2012 Free Software Foundation, Inc.
License GPLv3+/Autoconf: GNU GPL version 3 or later
<http://gnu.org/licenses/gpl.html>, <http://gnu.org/licenses/exceptions.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.

Written by David J. MacKenzie and Akim Demaille.

atrey@FNTSY MINGW64 /d/dev/ws/haskell/issues/network
$ autoreconf -i

atrey@FNTSY MINGW64 /d/dev/ws/haskell/issues/network
$ /d/bin/cabal.exe build
Resolving dependencies...
Build profile: -w ghc-8.10.5 -O1
In order, the following will be built (use -v for more details):
 - network-3.1.2.2 (lib:network) (first run)
Configuring network-3.1.2.2...
configure: WARNING: unrecognized options: --with-compiler
configure: loading site script D:/ghcup/msys64/etc/config.site
checking build system type... x86_64-w64-mingw32
checking host system type... x86_64-w64-mingw32
checking for gcc... D:\ghcup\ghc\8.10.5\lib\../mingw/bin\gcc.exe
checking whether the C compiler works... yes
checking for C compiler default output file name... a.exe
checking for suffix of executables... .exe
....
[28 of 28] Compiling Network.Socket.ByteString.Lazy ( Network\Socket\ByteString\Lazy.hs, D:\dev\ws\haskell\issues\network\dist-newstyle\build\x86_64-windows\ghc-8.10.5\network-3.1.2.2\build\Network\Socket\ByteString\Lazy.o )

The build ends with some warnings in all cases so i guess they are fine

[jneira]

I've tried with process and both versions of cabal works as intended. I think the cause is for process there are no config.guess and config.sub files.

I tested time with the same results

[jneira]

lgtm, all tests are ok I my end

[Mikolaj]

Yet another random fail from AppVeyor. I will rebase the branch and this will reset the tests and let's hope they pass.

[Mikolaj]

The CI finally finished successfully. Merged. @jneira: you rock. @emilypi: thank you for the review.

[Mikolaj]

@Mergifyio backport 3.6

[mergify]

Command backport 3.6: success

Backports have been created

• #7568 Set PATH_SEPARATOR=; when calling ./configure; fix #7494 (backport #7510) has been created for branch 3.6