Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

jenkins hsbencher 7.10 failure #93

Open
svenssonjoel opened this issue May 20, 2015 · 5 comments
Open

jenkins hsbencher 7.10 failure #93

svenssonjoel opened this issue May 20, 2015 · 5 comments
Assignees

Comments

@svenssonjoel
Copy link
Collaborator

http://tester-lin.soic.indiana.edu:8080/job/HSBencher/JENKINS_GHC=7.10.1,label=linux-soic/1017/console

This fails. Seems to happen during tests rather than install. Install does look a bit funny though.

@svenssonjoel svenssonjoel self-assigned this May 20, 2015
@tmcdonell
Copy link
Collaborator

Ryan just asked me to have a look at it. The log looks weird. I tried to
build manually and some of the components failed to install, so the log
might have picked up on the wrong error.

I just submitted a pull request that will fix the builds with 7.10
(although I didn't run the tests...)

On Thu, 21 May 2015 at 08:57 Joel Svensson notifications@github.com wrote:

http://tester-lin.soic.indiana.edu:8080/job/HSBencher/JENKINS_GHC=7.10.1,label=linux-soic/1017/console

This fails. Seems to happen during tests rather than install. Install does
look a bit funny though.


Reply to this email directly or view it on GitHub
#93.

@rrnewton
Copy link
Member

This error is usually from switching cabal version (from pre 1.22 to 1.22):

[cabalMethod] Failed to install bench1-0.1
 [cabalMethod]  [stderr] ghc: ghc no longer supports single-file style package databases (dist/package.conf.inplace) use 'ghc-pkg init' to create the database with the correct format.
 [cabalMethod]  [stderr] cabal: Error: some packages failed to install:

But wiping out workspaces should be sufficient to fix that. Some jobs are set to automatically wipe workspaces when the job fails.

@rrnewton
Copy link
Member

Ah, this job is wiping the workspace... but I think we're getting cabal skew even within the job.

On parfunc currently the default cabal executable is 1.20. But this job is set to explicitly use cabal-1.22 for GHC 7.10 support... and yet the hsbencher builds themselves just call "cabal", so somewhere in there I think we're using both 1.20 and 1.22 on the same directory.

(I had made the system wide default 1.22... but then ran into some other backwards compatibility problem that I'm forgetting right now. It would probably be good to have NO cabal in path and force jobs to select one explicitly or set up their paths.)

@rrnewton
Copy link
Member

Actually -- here's one quick fix. I propose a new policy for paths management.

The specific directories with GHC binaries:

  • ~/opt/ghc-7.10.1/bin
  • ~/opt/ghc-7.8.4/bin
  • etc

In addition to providing "ghc" and "ghci" we should also put appropriate versions of "cabal" (naked, no suffixes) in there. That way when we put ghc 7.10 in the path, cabal 1.22 automatically becomes the default. I did this just now for GHC 7.8 and 7.10. Let's see if it fixes that build. It's running now in build 1019.

@rrnewton
Copy link
Member

Yep, looks like it worked.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants