[Libre-soc-dev] daily kan-ban update 17may2021

[Luke Kenneth Casson Leighton]

On Tuesday, May 18, 2021, Lauri Kasanen <cand at gmx.com> wrote:


> BTW: it's painfully obvious all this is held together with bubblegum,
> not even tape, and barely runs even on the original devs' machines...


it does: i did say "rush-job" :)

the problem is that, actually, putting this very bluntly, nobody else in
the team is bothering to help keep thing up-to-date and stable.  being
honest i'm completely overloaded and overwhelmed and feel like i'm getting
absolutely no help at all from anybody.  it's not entirely true (it just
feels that way, to me) - both Cesar and Tobias are up-to-date and able to
contribute.

you're discovering these things because you're literally the first new
person in about a year to come forward and try to install things and run
them, Lauri.

also i was concerned about updating nmigen, because we are relying on a
version for the 180nm ASIC: turns out that, after investigation, i had in
fact updated local nmigen (12th april 2021) and i've just pushed that with
a forced master update
https://git.libre-soc.org/?p=nmigen.git;a=shortlog;h=refs/heads/master

i can confirm that those Simulation errors - which were something that was
occurring a year ago - are gone, with this commit
https://git.libre-soc.org/?p=nmigen.git;a=commit;h=d824795c2c7cb43dcbc8ed8fac6d309d77284913

as Jacob points out, there was a big overhaul of nmigen's Simulator code:
to keep the 180nm ASIC running i very deliberately chose *not* to update
nmigen until that had stabilised, and it has.

as it is a "forced master" update you may need to do "git checkout
d824795c2c
<https://git.libre-soc.org/?p=nmigen.git;a=commit;h=d824795c2c7cb43dcbc8ed8fac6d309d77284913>"
in the nmigen directory as a quick workaround.  if you've used "python3
setup.py develop" that *should* - if you are lucky - be all that's needed:
no need to re-run setup.py.

l.