Build instructions

[chris hermansen]

Now for my reply to Pietro's informative response.

On Thu, Jan 22, 2026 at 5:22 PM Pietro Monteiro <
pietro.gcc@sociotechnical.xyz> wrote:

> On Thu, Jan 22, 2026, at 1:25 AM, chris hermansen wrote:
> > Good day everyone,
> >
> > Yesterday, following Piero's advice, I cloned the source tree from GCC
> > rather than the forge and then ran a build (first time with non-forge
> > build).
> >
> > A few observations:
> >
> > - the build instructions at https://gcc.gnu.org/wiki/Algol68FrontEnd
> relate
> >   to working with the forge version; I see there is a similar set in
> >   gcc/gcc/algol68/README (once you have the source tree in place)
> >   but it lacks steps for building docs, running tests; could we offer a
> >   more full example on the wiki for using with the GCC source tree,
> >   including the URL for git clone?
>
> Besides the wiki A68 page, the documentation on building GCC also has a
> page on testing: https://gcc.gnu.org/install/


Super useful, thanks! I see there mentioned this build option:
--with-bugurl=urlShould we specify something here or leave the default?

>
>
> > - I seem to recall comments about using a parallel make process; is that
> >   a good idea?
>
> Yes, I usually do `make -j$(nproc)` or nproc-2 if I want to keep using my
> laptop while building GCC.
>

Thanks, that's very good to know. I will definitely use it the next time I
build.

>
> > - ../configure suggested another flag for me as I don't have i386 stuff
> >   installed on my machine, which was handy and maybe should be
> >   mentioned in the instructions
>
> Are you on a x86_64 system? I always build with multilibs enabled because
> that’s the easiest way to build/test 32-bit builds without using qemu or
> some
> remote setup.
>

Yes to x86_64.  I won't have anything that merits testing on any other
architecture for some time.


>
> > - maybe it was because I wasn't building with i386 but my build ran
> >   in much less time than my first build from the forge source tree.  I
> >   found this surprising, but no error messages hinting as to why.
>
> Did you disable bootstrap? A full bootstrap will build GCC at least 3
> times.
>

I don't believe so, at least not intentionally.  As I recall the last full
build I did from the forge source tree took ~ 1 day; the build I did from
the gcc git clone with no multi architecture took ~ 6 hours (these are very
rough times and I'm just going by memory; and without parallel make).

>
> > I would be happy to write a first draft to address what I mentioned
> > above, if anyone is interested, instead of just asking others to do it
> > for me.  Please let me know!
>
> Please do!
>

ok I will take a stab at writing something that combines the above info and
post it back to the list.


-- 
Chris Hermansen · clhermansen "at" gmail "dot" com

C'est ma façon de parler.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://gcc.gnu.org/pipermail/algol68/attachments/20260122/b60089b3/attachment-0001.htm>