[cfe-commits] [PATCH] Foundation for Transform registration infrastructure

[Manuel Klimek]

On Thu, Dec 20, 2012 at 3:39 PM, Vane, Edwin <edwin.vane at intel.com> wrote:

> I wasn't planning on dynamically loading transforms. Yesterday when I
> tried turning a transform into a static library to link into cpp11-migrate
> I was reminded that unused global variables like the RegisterTransform<...>
> vars won't get included in the link. My next goal is just to set up the
> build system to slurp through subdirectories adding all sources to the
> executable as one monolithic entity. I'm not sure how this is any less
> complex than using tablegen although I admit I don't have much experience
> with it.
>

Why tablegen? I'm still not sure why we need anything but simple classes
that can be instantiated...
Can you give examples of what you want to do that this wouldn't fulfill?

Thanks,
/Manuel


>
> -----Original Message-----
> From: Manuel Klimek [mailto:klimek at google.com]
> Sent: Thursday, December 20, 2012 2:43 AM
> To: klimek at google.com; Vane, Edwin
> Cc: cfe-commits at cs.uiuc.edu; gribozavr at gmail.com; silvas at purdue.edu
> Subject: Re: [PATCH] Foundation for Transform registration infrastructure
>
>
>   Sean's main point is my primary question: why do we need to load stuff
> dynamically?
>
>   I would expect to have a Transform class at some point, and have main
> basically look like:
>   if (Flag1)
>     Transforms.push_back(new T1);
>   ...
>   and then do something with the vector of transforms we've built - that
> wouldn't need any plugin infrastructure.
>
>   PS: regarding my rants - many people in llvm disagree with my positions
> here :) There's a lot of taste and personal preference to them, so if you
> agree with my reasoning, cool, but always feel free to argue if you don't
> agree :) I'm happy to learn new points of view
>
> http://llvm-reviews.chandlerc.com/D221
>
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20121220/100fe71a/attachment.html>