Re: Important: CVS move of preview-latex into AUCTeX completed.

Why didn't I see the original?

Jan-=C5ke Larsson <[email protected]> writes:

> Reiner Steib wrote:
>>> Since the move is now completed, also for dvipng, I will shortly
>>> give notice to SourceForge about this move to make sure we are not
>>> being in conflict with their policies.  The ideal solution I would
>>> want to have is to have the preview-latex site and pages at
>>> Sourceforge basically point to AUCTeX.  I'll see what parts would
>>> be reasonable to shut down there completely, and what parts of
>>> download area and CVS would remain, partly for archival reasons.
>> What is the current status?  I cannot find [preview/]dvipng in
>> AUCTeX's repository[1] and I can't check out dvipng from
>> sourceforge[2].
>
> dvipng has never had its CVS archive at sf, but now has one at
> savannah.nongnu.org:/cvsroot/dvipng
>
>> Should I list my changes to preview/dvipng/*, too?
>
> That should not be necessary.

It would be wrong for your assignment to AUCTeX.  Please don't.  The
only remote possibility of any meaningfulness would be if dvipng were
to be folded into the AUCTeX project proper at some point of time.

I have to admit that even though dvipng is a tailor-made race horse,
I'd like to avoid the can-of-worms of actually including binaries with
AUCTeX: lacheck has been folded out partly for that reason IIRC.

However, I think by now it would be reasonable to enable it if it is
available.  Checking for its presence at configure time and record
this in preview-image-type has a disadvantage: we'd have to make the
AUCTeX RPM and other packages depend on dvipng to get consistent
results.  It is probably more robust to do the check at run time: if
the first call to dvipng fails because it is not found, disable it
altogether for the session.

> Unless dvipng is to become part of the GNU project, too. And I
> haven't given that any serious thought yet.

Well, changes and contributions should be tracked and attributed
_always_.  Everything else is asking for trouble.  Copyright
assignments, however, are pointless before they are required from the
FSF.  Being a GNU project would not necessarily mean that: usually
Stallman decides whether it should.  If I remember correctly, dvipng
was derived from Nelson Beebe's drivers in some manner, though I might
be mistaken, so assigning it to the FSF would not be easy, anyway.

GNU projects usually have some strategic purpose.  I think that this
would become more relevant if some application coupled dvipng more
tightly into Emacs, for example (I really like the idea of using
dvipng's follow mode in connection with a continuing TeX process for
visualizing the "calc" symbolic calculator).  But since it would never
be standalone but would always require the background of a complete
TeX system (at the very least the fonts), the motivation of, say,
bundling dvipng with Emacs would be low.

Whatever.  Keep track of your contributions and take your time
thinking about applying for GNU status.  The main advantage for dvipng
would be somewhat nicer web addresses and downloads from ftp.gnu.org.

--=20
David Kastrup, Kriemhildstr. 15, 44793 Bochum