Re: [PATCH 3/6] gdbsupport: use dynamic partitioning in gdb::parallel_for_each

[Simon Marchi <simark@simark.ca>]

On 2025-07-03 15:23, Tom Tromey wrote:
>>>>>> "Simon" == Simon Marchi <simark@simark.ca> writes:
> 
> Sorry for the delay on this.  My time is split these days and so I'm not
> keeping up with gdb quite as frequently.
> 
> Simon> I have thought about this problem some more, and I discussed this with
> Simon> my team.  I still think that the functor approach is the only one that
> Simon> relatively cleanly meets the various requirements:
> 
> Simon>  - be able to run a per-worker initialize step
> Simon>  - be able to keep some per-worker data for the duration of the for-each
> Simon>  - be able to run a per-worker finalize step
> 
> IMO let's just go with your original plan.  It seems clear &
> straightforward.  I don't like the wordiness so much but I tend to think
> the other alternatives are less clear.

It is a bit more chatty, but I find it easier to understand, actually.
All the code and data used by the worker thread is in that class, wells
separated from the "main" path.

I will send an updated v2 then.

> This seems like another area that coroutines would clear up.  I have
> high hopes for those removing a lot of extraneous classes and turning
> them into local variables.  No idea when we can adopt C++20 though.

I'd be curious to see how that would look like for our parallel
for-each.

Simon