Add hidir setting - master branch#3968
Conversation
|
@christiaanb, thanks for your PR! By analyzing the history of the files in this pull request, we identified @dcoutts, @hvr and @ezyang to be potential reviewers. |
|
Is this ready for review or still WIP? |
|
This is ready for review |
|
|
||
| .. option:: --hidir=dir | ||
|
|
||
| Interfaces of libraries are installed here. |
There was a problem hiding this comment.
I'd expand this section a little bit to explain that normally hidir is the same as libdir <\> libsubdir, except on some broken systems where we must set libdir to "".
|
Looks OK to me. Just add a changelog note and this will be good to go. |
|
Looks like there are failing tests. |
c931357 to
84f5e3b
Compare
|
Yes, the failing test was due to https://ci.appveyor.com/project/ezyang/cabal/build/1.0.1740, however, that seems to be fixed in the latest master. I've rebased, everything should go through. |
|
I'm not sure your analysis is correct, but let's see. |
|
Yep, still fails. Looks like the problem is that we're not filtering out |
|
@christiaanb Look at |
Add a setting to indicate the directory for .hi interface files
84f5e3b to
dc16f19
Compare
|
Seems to be still not fully fixed. |
dc16f19 to
3a3b99c
Compare
|
OK, I need help... I thought that 3a3b99c ensures that the |
|
I'll look into this later today or tomorrow. |
c00ffeb to
c051c21
Compare
|
OK, I figured it out, all tests are green now. |
|
Nice. Are we OK with old Setups putting all .hi files in |
|
@23Skidoo No... that shouldn't happen. Although I see how my patch does that sigh Although I have no idea how to fix that properly.
|
|
I think that you have to refactor the |
|
Ugg. @23Skidoo do you recall how we used to make this work with ELF .so libs? I'm pretty sure they used to go into the $libdir not $libdir/$libsubdir. How did that work without us having a separate hidir? |
|
I want to note that I'm not too happy with c0ab0e8 either. But I'm feeling a bit under the weather, and I only have time to work on this during the evening, and there's the pending GHC 8.0.2 release. I tried to refector But that just turned into a big mess. I guess I could revert c0ab0e8 and then we'll just end up putting |
|
@dcoutts Not sure about that, but I've noticed that we have a Also there's this comment by @ezyang in D.C.Install:
|
Yeah, bad idea. If you want to do something like that, just do it for the particular field. (Have not been following convo, don't know the global picture.) |
|
Given the problem of copying |
|
The new PR is #3979 |
This is the version of #3955 based against master (instead of v1.24.0.0)
This pull request:
.dylib/.a) in a single shared directory in order to solve https://ghc.haskell.org/trac/ghc/ticket/12479