llvm dev - Jan 2015 - [LLVMdev] RFC: Could we move TargetLibraryInfo into lib/Analysis?

While this is target information, it isn't actually part of the target
today and doesn't use any part of the target code. Maybe some day we can
use a target hook to initialize the data structures, but even then I feel
like this really belongs in the core analysis library.

Would there be any objections to just moving it into Analysis?

Doing this would remove a seriously ugly layering violation we have today.
I believe this would remove the last dependency on lib/Target by
lib/Analysis and the largest set of dependencies on lib/Target by
lib/Transforms.

If there is general agreement, I will make it so.
-Chandler
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20150114/8e78f2af/attachment.html>

On Wed, Jan 14, 2015 at 12:03 PM, Chandler Carruth <chandlerc at
gmail.com> wrote:
> While this is target information, it isn't actually part of the target
today
> and doesn't use any part of the target code. Maybe some day we can use
a
> target hook to initialize the data structures, but even then I feel like
> this really belongs in the core analysis library.
So I'm curious: do you have an idea of where better to put this
hypothetical hook? Since it's mostly OS-dependent, it doesn't seem
very fit for the way we do targets.
> Would there be any objections to just moving it into Analysis?
>
> Doing this would remove a seriously ugly layering violation we have today.
I
> believe this would remove the last dependency on lib/Target by lib/Analysis
> and the largest set of dependencies on lib/Target by lib/Transforms.
>
> If there is general agreement, I will make it so.
FWIW I agree, on all counts.

-Ahmed
> -Chandler
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>

On Wed, Jan 14, 2015 at 9:46 AM, Ahmed Bougacha <ahmed.bougacha at
gmail.com>
wrote:
> On Wed, Jan 14, 2015 at 12:03 PM, Chandler Carruth <chandlerc at
gmail.com>
> wrote:
> > While this is target information, it isn't actually part of the
target
> today
> > and doesn't use any part of the target code. Maybe some day we can
use a
> > target hook to initialize the data structures, but even then I feel
like
> > this really belongs in the core analysis library.
>
> So I'm curious: do you have an idea of where better to put this
> hypothetical hook? Since it's mostly OS-dependent, it doesn't seem
> very fit for the way we do targets.
>
No idea at all, and I completely agree with you.

Ultimately, I think the word "target" here is misleading. It means
target
in the since of a cross compilation target, but it has little to no
relevance to LLVM targets. As you mention, it is an OS and runtime library
availability issue.

Anyways, we don't need a hook today, and if we do at some point in the
future we can figure out how to add one.

>
> > Would there be any objections to just moving it into Analysis?
> >
> > Doing this would remove a seriously ugly layering violation we have
> today. I
> > believe this would remove the last dependency on lib/Target by
> lib/Analysis
> > and the largest set of dependencies on lib/Target by lib/Transforms.
> >
> > If there is general agreement, I will make it so.
>
> FWIW I agree, on all counts.

Awesome. Will make it so. (Especially now that the release is branched.)
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20150114/b180a1da/attachment.html>