Re: [PATCH v2] Clarify Inline::C dependency (optional on Linux, required elsewhere)

[Eric Wong <e@80x24.org>]

Štěpán Němec <stepnem@smrk.net> wrote:
> On Sat, 02 Sep 2023 18:50:15 +0000
> Eric Wong wrote:
> 
> > Štěpán Němec <stepnem@smrk.net> wrote:
> >> -Numerous optional modules are likely to be useful as well:
> >> +Numerous other modules are likely to be useful as well:
> >
> > What is the reasoning for this change?
> > "optional" is an important point to state, IMHO.
> > Using "other" is more ambiguous, I think.
> 
> We've now established that at least Inline::C is not optional on all
> systems.  Is that really the only case?  (I mean, some of the modules
> have "optional" again in their respective parentheticals: is that just
> redundancy, or does it imply that the others are less optional?)  Even
> if it _is_ the only case, isn't saying "optional" misleading?

Perhaps the per-module "optional, for .." statements should be
"only for ...".

Aside from the new WIP -cindex; all public-inbox-* tools should
all work fine without Inline::C.  lei isn't likely running on
most public-facing servers.

In fact, all the components of public-inbox are optional
depending on which pieces you want to use it for.

The original public-inbox-* stuff (mda/learn/watch/WWW/httpd/CGI)
should work w/o SQLite for v1 inboxes, even (I need to test before
releases).

IOW, I'm trying to keep installations that were configured
and setup in back in 2014 working with no new dependencies.

Honestly, I wish I could tell users a C compiler is required to
make my life easier (heck, Varnish requires one); but gcc and
clang are both gigantic; and tcc (while active) hasn't had a
release in ages.

Thanks.