Re: implement .so

[Ingo Schwarze <schwarze@usta.de>]

Hi Joerg,

Joerg Sonnenberger wrote on Sun, Oct 24, 2010 at 06:49:45PM +0200:
> On Sun, Oct 24, 2010 at 06:40:57PM +0200, Ingo Schwarze wrote:

>> not only Xenocara (i.e. OpenBSD X.org), but various ports as well
>> use .so.  I'm not thrilled about that, i think .so is useful for
>> general-purpose typesetting, but rather dumb in a manual formatter.
>> Yet, being even less thrilled about the prospect to rewrite the
>> X.org build system or about patching lots of ports, i made up my
>> mind to implement .so.

> I don't think any patching itself has to be done.
> There is already a tool called soelim to expand them statically.

I fear we can't use that.
We are planning to install unformatted manuals in the future
because mandoc is so fast that it can format the manuals on
the fly, even on the VAX.
Having the source has quite a few advantages.

Installing soelim and running that in a pipe with mandoc
does not look sane.

Running soelim at build time and installing multiple copies
of the same page looks even less sane.
That would be waste of space, bandwidth...

Converting .so to hard links would be an option if we were talking
about just one build system.  But ports have all kinds of different
build systems.

> Remember that we discussed .so in Rostock?
> The behavior is not exactly sane.

Yes, i don't say i like it.

> So from a maintainance perspective, I think it is much easier to
> just handle the 99% of easy cases in ports/pkgsrc and patch the
> few remaining cases.

Hm.
I don't see how that could work.

Yours,
  Ingo
--
 To unsubscribe send an email to tech+unsubscribe@mdocml.bsd.lv