02:14 <_whitenotifier-8> [amaranth] whitequark commented on issue #1622: Spaces in names generates invalid `RTLIL` - https://github.com/amaranth-lang/amaranth/issues/1622#issuecomment-3138344751

03:02 <whitequark[cis]> <vup> is it intentional, that `data.Const` returns a boolean when comparing with another `data.Const`, while `_ast.Const` returns a `Operator("==", ...)`?...

03:02 <whitequark[cis]> yes

03:03 <whitequark[cis]> Value.__eq__ always returns another Value, we had an RFC that would change that and it was rejected

03:03 <whitequark[cis]> <vup> Furthermore, is there any reason, `data.Const` is not hashable?

03:03 <whitequark[cis]> I think that is an omission, iirc it's immutable

08:10 <_whitenotifier-8> [amaranth] rroohhh opened pull request #1623: lib.data.{Field, Layout, Const}: add __hash__ - https://github.com/amaranth-lang/amaranth/pull/1623

08:10 <vup> whitequark: do you have a link to the RFC?

08:13 <_whitenotifier-8> [amaranth] codecov[bot] commented on pull request #1623: lib.data.{Field, Layout, Const}: add __hash__ - https://github.com/amaranth-lang/amaranth/pull/1623#issuecomment-3138978426

08:22 <whitequark[cis]> <vup> whitequark: do you have a link to the RFC?...

08:22 <whitequark[cis]> not offhand, search for closed rfcs authored by me

08:23 <_whitenotifier-8> [amaranth] whitequark reviewed pull request #1623 commit - https://github.com/amaranth-lang/amaranth/pull/1623#discussion_r2244701816

08:25 <_whitenotifier-8> [amaranth] rroohhh reviewed pull request #1623 commit - https://github.com/amaranth-lang/amaranth/pull/1623#discussion_r2244709643

09:43 <_whitenotifier-8> [amaranth] whitequark reviewed pull request #1623 commit - https://github.com/amaranth-lang/amaranth/pull/1623#discussion_r2244886655

10:20 <_whitenotifier-8> [amaranth] zyp reviewed pull request #1623 commit - https://github.com/amaranth-lang/amaranth/pull/1623#discussion_r2244968386

10:21 <_whitenotifier-8> [amaranth] whitequark reviewed pull request #1623 commit - https://github.com/amaranth-lang/amaranth/pull/1623#discussion_r2244970122

10:34 <vup> zyp: how do i put `@functools.cache` on it without infinite recursion due to it calling hash itself? Is there a way I am missing to make it ignore the function arguments?

10:44 <zyp[m]> oh, duh, I didn't realize it'd call hash(self), but of course that makes sense…

10:47 <vup> well, I cannot come up with something better than https://paste.debian.net/1388919/ I think. Doing it in the constructor, would require the subclasses to call super().__init__()... Also it would be nice to avoid having the `self._hash` pollute the useable names of the subclass, but I am not sure how to do that

10:51 <zyp[m]> if you use a double underscore prefix, it'll be mangled in subclasses to avoid collision

10:51 <zyp[m]> not entirely sure how that interacts with hasattr though

10:55 <vup> ah right totally forgot about that

10:59 <_whitenotifier-8> [amaranth] rroohhh reviewed pull request #1623 commit - https://github.com/amaranth-lang/amaranth/pull/1623#discussion_r2245059725

11:03 <_whitenotifier-8> [amaranth] whitequark commented on pull request #1623: lib.data.{Field, Layout, Const}: add __hash__ - https://github.com/amaranth-lang/amaranth/pull/1623#issuecomment-3139486399

11:46 <_whitenotifier-8> [amaranth] whitequark commented on pull request #1623: lib.data.{Field, Layout, Const}: add __hash__ - https://github.com/amaranth-lang/amaranth/pull/1623#issuecomment-3139644858

17:45 <AhmedCharles[m]> I just realized that amaranth/yosys are doing dead code (circuit?) elimination on my riscv core based on which instructions exist in the memory, so in order to compare the size of the design in LUTs, I need to ensure that the memory contains each type of instruction, even if they aren't executed. At least it's not smart enough to figure out that the riscv instructions are dead code, yet.

17:46 <whitequark[cis]> yosys does that by first turning your ROM into a switch

19:20 <grabhanem[m]> I assume it won't do that if you have executable ram?