diff options
author | Peter Scheibel <scheibel1@llnl.gov> | 2019-10-23 16:50:51 -0700 |
---|---|---|
committer | Todd Gamblin <tgamblin@llnl.gov> | 2019-10-23 16:50:51 -0700 |
commit | f2ddffb8400934d6b2e975d359be22c1997d6863 (patch) | |
tree | 24d43025376377d93de365e8fcee27f79791cd56 /.gitlab-ci.yml | |
parent | bd7265ed5226078fd13d2fe55164a5e53f3193e3 (diff) | |
download | spack-f2ddffb8400934d6b2e975d359be22c1997d6863.tar.gz spack-f2ddffb8400934d6b2e975d359be22c1997d6863.tar.bz2 spack-f2ddffb8400934d6b2e975d359be22c1997d6863.tar.xz spack-f2ddffb8400934d6b2e975d359be22c1997d6863.zip |
bugfix: web.push_to_url should not format the local path that is passed to it (#13408)
fd58c98 formats the `Stage`'s `archive_path` in `Stage.archive` (as part of `web.push_to_url`). This is not needed and if the formatted differs from the original path (for example if the archive file name contains a URL query suffix), then the copy fails.
This removes the formatting that occurs in `web.push_to_url`.
We should figure out a way to handle bad cases like this *and* to have nicer filenames for downloaded files. One option that would work in this particular case would be to also pass `-J` / `--remote-header-name` to `curl`. We'll need to do follow-up work to determine if we can use `-J` everywhere.
See also: https://github.com/spack/spack/pull/11117#discussion_r338301058
Diffstat (limited to '.gitlab-ci.yml')
0 files changed, 0 insertions, 0 deletions