[PATCH] D43093: [FastISel] Sink local value materializations to first use

[Zachary Turner via llvm-commits]

I would argue that It’s a useful for debuggers in general that want to be
able to implement a feature to move the instruction pointer to the next
statement without breaking the state of the program. So I wouldn’t look at
this so much as catering to the whims of a specific debugger as I would
enabling debuggers in general to work more effectively. And at O0, the
entire point should be a good debugging experience imo.
On Tue, Feb 27, 2018 at 1:37 PM Matthias Braun via Phabricator <
reviews at reviews.llvm.org> wrote:

> MatzeB added a comment.
>
> > Now, should we go to such length to make the debug information be
> friendlier to some debugger, I am not sure, but I let you decide.
>
> Definitely not! That debugger constraint feels like it is designed for
> ancient compilers producing code and register allocating for one C
> statement at a time.
>
> Luckily this change makes sense independently of the debugger used IMO.
>
>
> https://reviews.llvm.org/D43093
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180227/c55389f4/attachment.html>