Mail Archives: djgpp/2011/11/14/12:45:18
Hi,
(deleted LONG rant, email me if curious)
On Nov 14, 1:36 am, "Stephen J. Turnbull" <turnb DOT DOT DOT AT sk DOT tsukuba DOT ac DOT jp>
wrote:
> Eli Zaretskii writes:
>
> > Again, the point is that to build a package, one should use only a
> > small and portable set of utilities.
>
> "Should be portable", yes, but "should be small" is no longer obvious
> for typical applications, especially on "modern" desktop platforms.
> It's true for those packages needed to bootstrap a system, but after
> that DRY ("don't repeat yourself") becomes more important to many,
> perhaps most, developers.
The computer industry is very short-lived (except for copyright and
patents, ironically) and changes ridiculously quickly. Thus nothing
lasts very long (and is sometimes against common interest). A lot of
the hype is to get people to upgrade. In some ways, it all thrives on
that constant upgrade policy, constantly making things better.
Unfortunately, anything that isn't top tier is relegated to
"legacy" (vs. "modern"), which is a stigma that affects various things
negatively. And people usually ignore anything labeled legacy, even
when it supports the same functionality. Often "standards" themselves
are crippled or outdated or even abandoned due to such reasons.
In other words, it's a tug of war between progress / innovation and
useful / functional / stable / good enough. (Or maybe business vs.
freedom vs. art.)
> It could be that the package in question qualifies as a bootstrap
> package, or as a member of the small & portable set of utilities used
> in building such packages. However, in many cases what we see is
> people complaining about lack of portability of a program developed in
> a tool-rich environment for use in the tool-rich environment. While I
> sympathize with DJGPP developers and embedded developers who may not
> work in such environments, as suggested earlier that requires a
> volunteer willing to do the portability work.
Sometimes working patches and builds are rejected for purely
political, polemic, or philosophical reasons or even laziness. It's
not always for lack of volunteers. You have to admit that "DOS" is not
a trendy word these days. And yes I speak from experience (more from
persistence than anything as I don't code very much).
> > The GNU Coding Standards require that.
>
> More honored in the breach than the observance these days.
GNU officially only codes for the GNU system (which I guess? means
Linux and not Hurd these days). But they do also cater to popular
platforms ("Woe32" / Windows), which most users want, as well as those
(seemingly obscure) that have commercial support.
- Raw text -