llvm dev - Oct 2020 - Manipulating DAGs in TableGen

I was wondering today whether the TableGen DAG support would be more useful if
you could manipulate DAGs a little more easily. Right now, you can:

* Specify a DAG with (...)

* Catenate DAGs with !con().

* Construct a DAG with !dag().

* Iterate over a DAG with !foreach().

* Get and set the DAG operator with !getop() and !setop().

What if you could also get and set a DAG's operands individually, by
position or by $xxx name? Would that add enough power/convenience that you would
build more interesting DAGs in TableGen, or is it just easier to do such
building in the backends?

Yeah, this is definitely useful in my experience. I had the same idea a few
days ago but it sidetracked in my ToDos.

IMHO, it is considerable effort though.

On Fri, Oct 9, 2020, 11:38 PM Paul C. Anagnostopoulos via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> I was wondering today whether the TableGen DAG support would be more
> useful if you could manipulate DAGs a little more easily. Right now, you
> can:
>
> * Specify a DAG with (...)
>
> * Catenate DAGs with !con().
>
> * Construct a DAG with !dag().
>
> * Iterate over a DAG with !foreach().
>
> * Get and set the DAG operator with !getop() and !setop().
>
> What if you could also get and set a DAG's operands individually, by
> position or by $xxx name? Would that add enough power/convenience that you
> would build more interesting DAGs in TableGen, or is it just easier to do
> such building in the backends?
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20201009/2cc17006/attachment.html>

At 10/9/2020 02:22 PM, Madhur Amilkanthwar wrote:
>Yeah, this is definitely useful in my experience. I had the same idea a few
days ago but it sidetracked in my ToDos.Â
>
>IMHO, it is considerable effort though. 
It is some work, for sure.

Let me post a modest proposal here in the next few days. I'd appreciate
comments and suggestions. The idea is to make it easier to manipulate complex
DAGs in TableGen.