Re: [Maxima-discuss] Texinfo version, maxima.info build target, was: Help paths (was: Re: Maxima help can use browser now)

[Raymond T. <toy...@gm...>]

On Tue, Jul 19, 2022 at 9:11 AM Robert Dodier <rob...@gm...>
wrote:

> On Mon, Jul 18, 2022 at 6:05 PM Raymond Toy <toy...@gm...> wrote:
>
> >> About the makeinfo version, it seems like that should be enforced by
> >> ./configure (unless --enable-build-docs=no).
>
> > Yeah, I thought about it, but I didn't know how to get that enforced.
> At least not without hacks on pattern matching the output from makeinfo
> --version.
>
> Well, since it appears to be pretty important to get the right Texinfo
> version, it seems like we're kind of obligated to do it. configure is
> a huge collection of hackery, so one more won't stand out.
>

I'll see what I can do.  Or look into allowing 6.7 instead, but I'd prefer
not to.

>
> I've pushed 6286fee which fixes a problem (@menu stuff) which
> prevented makeinfo from succeeding. Now make in doc/info succeeds, and
> maxima-index.lisp is created, and describe works again.
>

Oops.  I'm pretty sure I asked texinfo mode to update all menus and nodes.
Maybe I forgot.  But it seems odd that Binomial wasn't included, but the
later ones are.

>
> It seems odd to me that makeinfo failed, but it didn't halt the build
> -- I would have expected make to terminate with nonzero status. Make
> happily plowed ahead and executed build_index.pl, which didn't notice
> maxima.info doesn't exist. I guess build_index.pl could notice the
> nonexistent maxima.info, but I think the more important thing is for
> make to notice that makeinfo failed.
>

I think this is because we use makeinfo --force, which ignores any errors,
so presumably the return code is 0 still, making everything look like it
succeeds.

I still have some errors in makeinfo.  I've been meaning to look into
fixing them so that we don't have to use force anymore.

>
> best,
>
> Robert
>


-- 
Ray