<_whitenotifier-5>
[amaranth-lang/amaranth-lang.github.io] whitequark ea2dedd - Deploying to main from @ amaranth-lang/playground@68fb2a0c360ca4d84c0749e188601c7bce7b690d 🚀
<whitequark[cis]>
<galibert[m]> "That's a beautiful comment you..." <- I was... unhappy
<galibert[m]>
You seemed... opiniated at that point in time
<whitequark[cis]>
I wrote that a few days ago lol
<galibert[m]>
heh, didn't check the timing
Wanda[cis] has joined #amaranth-lang
<Wanda[cis]>
heh
<Wanda[cis]>
I stand by this, by the way
<Wanda[cis]>
(that comment in large part came from my frustration too)
<galibert[m]>
shared frustatration then
<Wanda[cis]>
you have no idea how much time I have wasted on dealing specifically with the init attribute
<galibert[m]>
I believe you
<galibert[m]>
is that one of the differences between yosys rtlil and amaranth rtlil?
<Wanda[cis]>
wtf is amaranth rtlil
<galibert[m]>
Errr no, I meant IR, duh
<Wanda[cis]>
oh. well, yes.
<galibert[m]>
Somehow I equated IR and rtlil
<Wanda[cis]>
oh gods no
<galibert[m]>
In fpga land the gods have left in disgust a long time ago
<galibert[m]>
Isn't the interface between amaranth and yosys rtlil though?
<Wanda[cis]>
hm?
<Wanda[cis]>
well
<Wanda[cis]>
this is the file that converts amaranth NIR into RTLIL, yes
<galibert[m]>
amaranth is talking ascii rtlil to yosys, right?
<Wanda[cis]>
yes
<galibert[m]>
so #1100 is about creating some abstract rtlil-representing objects with ascii conversion in them, and generate those from nir and not directly?
<Wanda[cis]>
something like that, yes
<galibert[m]>
makes sense
<Wanda[cis]>
basically we want to do scene graph rendering of RTLIL, instead of immediate mode rendering
<galibert[m]>
hehe yeah
<Wanda[cis]>
the current code has several places with horrible hacks because basically what we wanted to do was add some attributes to a wire after the fact
<Wanda[cis]>
s/ wanted/_wanted_/, s/the/it/, s/fact/was first emitted/
<Wanda[cis]>
which is of course not possible in immediate mode
<_whitenotifier-7>
[amaranth-lang/amaranth-lang.github.io] github-merge-queue[bot] 3d437df - Deploying to main from @ amaranth-lang/amaranth@09029cdd914d550e924cecfb9aa1692a536b4161 🚀
<cr1901>
I've been trying to document an Amaranth class, and cross-ref to Amaranth docs, and have been wondering why even w/ intersphinx, I can't get :class:`amaranth:Signal` to work
<cr1901>
Your code works. I also have python in my intersphinx_mapping, and "{class}`python:zipfile.ZipFile`" works while "{class}`amaranth:Value`" doesn't. Do you have any idea why?
<cr1901>
shit... ignore that
<cr1901>
:class:`python:zipfile.ZipFile`* works, :class:`amaranth:Value`* does not
<cr1901>
The correct cross ref is :class:`amaranth:amaranth.hdl.Value` (marked up differently than :class:`amaranth.hdl.Value`, though they resolve to the same place)
<whitequark[cis]>
yep
<whitequark[cis]>
I don't think I even can supply shortcuts like that?
<cr1901>
Ahhh
<cr1901>
Is there a way to express a Signal with specific shape as part of docs (e.g. as part of the Attributes)? I know a particular attribute "foo" is of type Signal, but the user might want to know that foo is "a Signal with shape Bar".
<whitequark[cis]>
we should figure out a proper format for this eventually
<cr1901>
the :py:class: role has provisions for type hints/generics, but it's not clear to me that there's a standardized way to represent "type with instantiated generic parameters" or "in/out". The docs for crc.Processor look okay, but it's obvious that napoleon/sphinx didn't do any special markup w/ the bespoke format :P https://amaranth-lang.org/docs/amaranth/latest/stdlib/crc.html#amaranth.lib.crc.Processor