configure.ac: Export MAKECMD to build system
ClosedPublic

Authored by kgardas on Jan 13 2016, 5:13 PM.

Diff Detail

Repository
rGHC Glasgow Haskell Compiler
Lint
Automatic diff as part of commit; lint not applicable.
Unit
Automatic diff as part of commit; unit tests not applicable.
kgardas updated this revision to Diff 6183.Jan 13 2016, 5:13 PM
kgardas retitled this revision from to save detected make command into mk/config.mk for shake purposes.
kgardas updated this object.
kgardas edited the test plan for this revision. (Show Details)
kgardas added a reviewer: bgamari.
bgamari added inline comments.Jan 14 2016, 7:25 AM
mk/config.mk.in
696 ↗(On Diff #6183)

I'm not sure I see why this is necessary. The shake build system doesn't use mk/config.mk. I think the configure.ac changes are sufficient.

kgardas updated this revision to Diff 6188.Jan 14 2016, 8:30 AM
kgardas edited edge metadata.

do not set make command in mk/config.mk.in (not necessary for shake)

kgardas marked an inline comment as done.Jan 14 2016, 8:31 AM

@bgamari you are right about it so I removed that.

bgamari added inline comments.Jan 14 2016, 8:49 AM
configure.ac
1125

Perhaps we should fail in this case?

bgamari requested changes to this revision.Jan 15 2016, 6:10 AM
bgamari retitled this revision from save detected make command into mk/config.mk for shake purposes to configure.ac: Export MAKECMD to build system.
bgamari updated this object.
bgamari edited edge metadata.
bgamari edited edge metadata.

@kgardas, this seems to have failed validation. I've restarted the validation run to ensure this wasn't a fluke but I suspect there is something wrong here.

This revision now requires changes to proceed.Jan 15 2016, 6:11 AM

@bgamari failed on ghc-stage1 running out of memory. Can't see how this may be related to the proposed change...

configure.ac
1125

I've tried to follow current behaviour. That is to just print warning. If you like to fail hard in this case, then we should do that before generating all config files and that means at the place of checkMake380 probably.

@bgamari failed on ghc-stage1 running out of memory. Can't see how this may be related to the proposed change...

Indeed it's not. Harbormaster is unfortunately just a bit fragile. We are working on fixing this.

@thomie, do you see any reason not to fail here?

It doesn't really matter, warning or error.

I would just delete that check. It's for a bug in a version of make that nobody uses (2002).

@thomie Hold on, gmake 3.80 is distributed with Solaris 10 up to these days IIRC and even there is nearly zero chance someone will update their GHC to 8.x on this system, still if the warning hurts nobody I would rather keep it there.

thomie accepted this revision.Jan 15 2016, 10:15 AM
thomie added a reviewer: thomie.

Ok, thanks.

This revision was automatically updated to reflect the committed changes.