fenics team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Tue, Apr 23, 2013 at 12:11:44PM +0200, Johannes Ring wrote:
> On Mon, Apr 22, 2013 at 9:22 PM, Anders Logg <logg@xxxxxxxxx> wrote:
> > FFC reference data are now stored here:
> >
> > https://bitbucket.org/fenics-project/ffc-reference-data
> >
> > The following two scripts are used to download / upload the data:
> >
> >   ffc/test/regression/download-references
> >   ffc/test/regression/upload-references
> >
> > The download script is not yet run automatically as part of running
> > the test suite. My suggestion would be to just call the download
> > script at the beginning of test/regression/test.py. Johannes, if
> > you think this can work out for the buildbots please add it.
>
> The buildbot does not have access to
> ssh://git@xxxxxxxxxxxxx/fenics-project/ffc-reference-data.git, so it
> won't work unless this is changed to
> https://bitbucket.org/fenics-project/ffc-reference-data.git. I did
> this and tried a run on the buildbot:
>
>
>http://fenicsproject.org:8010/builders/ffc-master-quantal-amd64/builds/13/steps/test/logs/stdio

Looks good.

> I used the following patch:
>
> http://fenicsproject.org:8010/builders/ffc-master-quantal-amd64/builds/13/steps/git/logs/patch

Why do you use the DATA_REPO_SSH in the first case here?

  if [ ! -d $DATA_DIR ]; then
       echo "Cloning reference data repository"
  -    git clone $DATA_REPO
  +    git clone $DATA_REPO_SSH
  +    if [ $? -ne 0 ]; then
  +    git clone $DATA_REPO_HTTPS
  +    fi
  else

Shouldn't it be the DATA_REPO_HTTPS in both cases?

I assume the logic is to use ssh in the upload script and https in the
download script.

--
Anders