CI failure: can't find include file rts/fs.h #554
Comments
I haven't yet looked at this but I have a feeling that this is caused by a missing dependency on These files are copied since #533. Also, this code may be relevant: Lines 38 to 58 in 3a68f11
Or not :) |
The bug is the |
HACK: |
failed with another (probably unrelated) bug: #559 |
@Mistuke @snowleopard any idea on how to understand the HACK and fit it in our code? |
I don't quite understand why those includes are treated differently than the others. They're copied in tree way before any dependency analysis is done. But we probably don't need to install them anyway. Try in the GHC checkout removing Note that it's not the building that fails, it's the package installing phase. I don't have a Linux machine handy atm. but if you can't modify the tree easily I'll go set up a vm. |
I am with @Mistuke here: I don't quite understand why But I'm getting more convinced that I'm right in suspecting that I'll try to have a look at this issue tonight. |
I tried and it worked. |
I've added a workaround to fix the -- TODO: Get rid of this workaround.
-- See https://github.com/snowleopard/hadrian/issues/554
root -/- buildDir rtsContext -/- "rts/fs.h" <~ return "rts"
root -/- buildDir rtsContext -/- "rts/fs_rts.h" <~ return "rts" The right way to fix this could be to add @izgzhen @Mistuke What do you think? In the meanwhile, the current workaround works, so I'm lowering the priority of this issue. |
looks good! |
Yes that's good, though looking at it again. I probably don't care about
having these installed. As the user can't use them easily anyway since they
need to know the right name space. I will commit a change removing these
from install once I get home.
Sorry for the extra work.
…On Thu, Apr 12, 2018, 03:02 Zhen Zhang ***@***.***> wrote:
looks good!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#554 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABH3Kc_4WBicbhLMrGbBaccFGafuR9akks5tnrWhgaJpZM4TExys>
.
|
I still saw them in error msg on my machine building the latest commit though :( no yet investigating into it |
Commit ghc/ghc@111556f removes these files from |
Actually I've had to revert this, now I remember why they were there to begin with. The reach-ability analysis forces them to be. Because they're reachable from an exported header. This is all very annoying. The exported headers are not useful. In fact you can't even call them so moving them to |
sigh well that's annoying. the CI failure was unrelated. I put up https://phabricator.haskell.org/D4591 for review to run it through the mac osx builder. if it's green I'll push the removal again. Sorry for the delay. |
ghc/ghc@5417c68 done. the CI was clean so the failure was unrelated. You should be able to remove the work-around @snowleopard |
@Mistuke Great! I'll remove my workaround and will close this today. |
Done in the above commit. I think we can close this now. |
https://travis-ci.org/snowleopard/hadrian/jobs/361477751
https://travis-ci.org/snowleopard/hadrian/jobs/361477747
please leave a word if you are looking into this bug, thanks!
The text was updated successfully, but these errors were encountered: