Re: icg build instructions [was: Re: igc: trying to chase a crash]

["Gerd Möllmann" <gerd.moellmann@gmail.com>]

Óscar Fuentes <ofv@wanadoo.es> writes:

> Gerd Möllmann <gerd.moellmann@gmail.com> writes:
>
>> Andrea Corallo <acorallo@gnu.org> writes:
>>
>>> A README-IGC in the top-level dir SGTM (as a temporary solution before
>>> the merge at least).  I'm wondering if we could have some way to
>>> propagate this information also outside emacs.git as well 🤷
>>>
>>>   Andrea
>>
>> I've added a README-IGC, with what I remember. Maybe others could add
>> what they know to it.
>
> Thanks.
>
> AFAIK, mps is not packaged by Debian.
>
> For building it locally, possibly the easiest route is to clone
>
> https://github.com/Ravenbrook/mps.git
>
> (which was updated just today, BTW), compile with

Yeah, it appears they are up to something. I think I have 3 dozen mails
from Github from the last two or three days, from watching their repo.
Merging pull requests and so. Maybe they are preparing a new release or
something?

> cd code
> cc -O2 -c mps.c
> ar rvs libmps.a mps.o
>
> Then config the Emacs build with
>
> configure CPPFLAGS=-I/path/to/mps/includedirectory
> LDFLAGS=-L/path/to/mps/codedirectory --with-mps=yes <... rest of
> configure options..>
>
> This can have a variation for compiling&using the debug version of mps.
>
> The above is not tested with the current state of mps/emacs. I can try
> later and add those instructions to README-IGC.

That would be nice.

>
> Whe should decide if we direct users towards the optimized build (better
> engagement and real-case experience) or to the debug build (better bug
> reports but unusably slow in my experience).

Yes. Maybe users would be disappointed by the performance when running
with full debug? Don't know.