llvm : added version 3.8.0 #591
Conversation
| @@ -52,7 +52,7 @@ class Llvm(Package): | |||
| depends_on('cmake @2.8.12.2:') | |||
|
|
|||
| # Universal dependency | |||
| depends_on('python@2.7:') | |||
| depends_on('python@2.7:2.8') # Seems not to support python 3.X.Y | |||
There was a problem hiding this comment.
This can be 2.7:2.7. The range isn't half-open.
There was a problem hiding this comment.
@mathstuf Are you sure this will include anything that matches 2.7.X? I just tried and got:
==> Warning: There is no checksum on file to fetch python@2.7 safely.
Fetch anyway? [y/N]
Another issue is, I think, that somehow the preferred keyword got lost in the versioning. If I try :
spack install python
spack installs 3.5.1, even if in the python package I see :
version('2.7.11', '6b6076ec9e93f05dd63e47eb9c15728b', preferred=True)
There was a problem hiding this comment.
Ah, maybe because both sides are the same, it is as if there's no range. Yeah, I also saw 3.5.1 being pulled in, hence #592.
There was a problem hiding this comment.
If you just want it to match 2.7.x, depends_on('python@2.7') should work. If it doesn't it's a bug.
There was a problem hiding this comment.
And preferred was lost in PR #120. @mplegendre is going to add it back if I don't get to it first.
There was a problem hiding this comment.
I had tried @2, but that didn't try to match, but instead it tried to use version 2 exactly.
There was a problem hiding this comment.
@2.7 is @2.7.0. You want to allow @2.7.11 as well. I'd write @2.7:2.7.999.
There was a problem hiding this comment.
@2.7 should match 2.7, 2.7.x, and anything more specific, and @2 should match 2, 2.x, 2.x.y, etc. Is this not the behavior you're seeing? I'll check the tests.
I have wanted to put an exact match syntax in for a while, where @2.7. would match only 2.7 exactly (no additional specifiers) but I have not done that yet.
There was a problem hiding this comment.
When I used @2, it was trying to install exactly version 2.
@tgamblin I don't really understand why @eschnett update is mentioned here after pulling from upstream. If it is polluting in some way the history, I'll just cherry pick the changes and prepare another PR