> -----Original Message----- > From: Sam James <[email protected]> > Sent: 04 August 2026 23:42 > To: Kyrylo Tkachov <[email protected]> > Cc: Jeffrey Law <[email protected]>; Tamar Christina > <[email protected]>; [email protected]; > [email protected] > Subject: Re: [PATCH 1/2] genopinit: Distribute generated code across multiple > files > > Kyrylo Tkachov <[email protected]> writes: > > >> On 4 Aug 2026, at 20:34, Jeffrey Law <[email protected]> > wrote: > >> > >> > >> > >> On 8/4/2026 10:59 AM, Tamar Christina wrote: > >>> So I've been ignoring this since it's mostly Richard's thing and we seem > >>> to > disagree > >>> on design a lot. > >>> > >>> But in my opinion we don't want to keep adding split code to every gen* > file individually > >>> but concentrate that code in gensupport so ever gen* can use it. > >> I think that's a fair concern. We should have enough data on how to > >> split the gen* things derived from .md files to be able to see the > >> commonality as well as behaviorial differences we need to put in > >> place. So it's probably a good time to look at that refactoring. I > >> don't think anyone believes things are anywhere near optimal, but I > >> know for the ones Robin did last year, the affect was measured in > >> hours of wall clock time on RISC-V, so huge wins without even trying > >> too hard to be "optimal". > > > > Thanks for the background Tamar. I remember there was some activity in > > that area, but I had not done the homework tbh. I have a bit of time > > to dedicate in this area at the moment so I’ll see if I can come up > > with a sensible refactoring along these lines. > > I’ll hold off on committing these two patches for now, unless you want them > in in the meantime. > > Personally, I'd like to see them in if others are OK with it, because I > have some backports for older branches that I use for convenience when > bisecting, and something more incremental is easier to use there too. >
Well a patch putting the shared code in gensupport would be smaller :) > But this isn't very important if people prefer to wait. > Please don't hold up committing this on my account. My comments were merely to point out that the approach won't scale too well past this point and that there other smaller tweaks that could be made to the codegen of these files that would help the compiler out. Thanks, Tamar > > > > Kyrill > > > >> > >> Jeff
