Telcon: 2021 09 15
Peter Scheibel edited this page Sep 15, 2021
·
13 revisions
Weekly Spack meeting to be held on Wednesday September 15th 9am PST
- Todd Gamblin
- Greg Becker
- Peter Scheibel
- Kayla Butler
- Tammy Dahlgren
- Massimiliano Culpo
- Asher Mancinelli
- Brian Van Essen
- Mark Krentel
-
Brian Van Essen: long downloads for CUDA from Spack mirror: takes forever
- With
use_curl: True
, this takes 50 minutes - Just using wget (outside of Spack), takes 5 minutes
- With
-
When should Spack bail out after concretizing to a deprecated spec, and how hesitant should we be to deprecate usable specs?
- Issue: Spack concretizes to a deprecated version that it won’t build, but it keeps going until getting a messier error in the CI pipeline than we would expect.
- (CHOSEN) Proposal (Todd): add a config option to fail when trying to install deprecated specs
- Off by default (?)
- Example: openssl (commonly want to deprecate this)
- There was a pipeline that failed for a new PR, set to prefer openssl@1.1...k
- Before the PR, a new version of openssl was added
- Proposal (Greg): if a package isn't getting supported by the maintainers, we should still not deprecate
- This is mainly for security concerns
- Proposal (Massimiliano): don't immediately fail when trying to install deprecated specs. Still mark them as deprecated.
- Mark Krentel:
spack info
doesn't report deprecated versions at all. It would be useful to report these but perhaps with a note about them being deprecated. - Proposal (Mark): could we add an option to describe why a Package version is deprecated.
- Brian: want to deprecate all pre-1.0 versions of
lbann
(not necessarily fail if a user requests it) - Original PR (which may contain additional notes on usage): https://github.com/spack/spack/pull/20767
-
(TBC from last week) Discussion: Packages which require multiple build systems: how to handle them?
- Discussions on this:
- (Proposal based on using a new directive) https://github.com/spack/seps/blob/8dafadb4cae08275d168af964b3d168b4248977f/seps/sep-0002.md
- (Notes on trying an approach based on a
when
class decorator) https://github.com/spack/seps/issues/4
- Two approaches were covered: Greg thought there were issues with both; we will cover that next week.
- Discussions on this:
-
Discussion: Packages with multiple conflicting libraries - how to support them?
- For now, any such package installed by Spack advertises just one from the set (this ensures that everything is consistent where needed)
- https://github.com/spack/seps/pull/2
- Brian: opencv will try to install its own openblas (which may conflict with what Spack chooses)
- Mvapich2 installs its own hwloc
- (Peter) I think this is a separate but still interesting problem.
-
(For next week) What about packages that install some other package with them
- E.g. Mvapich2 installs hwloc with it
- Would it be good to allow a Spack package to describe these additional packages that "come along" with it
-
Discussion: long environment activations - are you having an issue with this?
- Some large envs might have this problem, e.g.: https://github.com/spack/spack/issues/25555
-
Pending SEP proposal from Chuck Atkins regardings discussion on a standard for shared/static libs, we will discuss that