[omega] [tex-live] ovp2ovf with VPL input

[Lars Hellström]

Peter Breitenlohner skrev:
> On Tue, 15 Sep 2009, Karl Berry wrote:
> 
>> As far as I'm concerned they are frozen now (and have been for some
>> time).  I don't plan to ask for more rebuilds unless there is some
>> really disastrous new bug report in the core programs.  And sorry, but I
>> don't count ovp2ovf as a core program :).

Well, then at least put a big sticker in the manpage or something 
saying "ovp2ovf appears to be broken in TeXLive 2009, we're working on 
fixing it, see http://... for further info" or some such.

> 
> Hi Lars,
> 
> moreover in cases where one could as well use VTtoVP & Co.

You mean VFtoVP? Problem is, I want to make an 8-bit VF that does a 
16-bit set_char, and VPtoVF screams if the argument of (SETCHAR) 
exceeds 255. ovp2ovf exposes more of the generality of the DVI format.

>> Lars, I think the best approach will be to take our time and debug it
>> properly.  It's no problem to compile binaries for you to use when the
>> time comes, regardless of whether we update TL.

*I* can build (from the web source) an ovp2ovf that does what I want; 
indeed, I've already done that. The problem comes when other people 
might want to repeat what I'm going to use it for...

>> Peter Breitenlohner has also been working on the source tree; I don't
>> know if he's attempted to deal with the parse errors.
> 
> Hi Lars,
> 
> both the C and the Web2C version are broken in many respects, in particular
> neither version properly handles tfm/vf files.

The only problem with the Web version that I've seen is that it lacks a 
method for the user to make it generate VF+TFM (set ofm_level=-1, in 
that case). Easy enough to make a command-line option for, or 
conditionalise by the input file suffix...

> In my private copy of the tree (for future developments) I have some fixes
> for the Web2C versions at least.  This, however, has to wait until after 
> the
> freeze of the TeXLive SVN tree.

\begin{off-topic}
   Is there somewhere a repository going back more than the 3 years
   of svn://tug.org/texlive? Last night I had a run-in with some
   code for big fonts (>256 chars) in dvips, and wanted to see how
   it had evolved, but lacked a history to investigate. (By accident,
   I discovered that some of it had been put in by a certain Peter
   Breitenlohner during the dvips&odvips unification in 2005, but it
   would be interesting to know to what extent older dvipses had such
   support too.)
\end{off-topic}

>> As I recall, Akira needed the C implementations of these tools, and that
>> is why we are using them, instead of the older web implementations.
> 
> It's just the other way around. Akira needed the Web2C implementations for
> some ofm_level=1 feature (CHARREPEAT) completely absent from the C version.

So there are so far no known examples of the C version (2.0) doing 
anything that the Web version (1.11) can't do just as well or better?

Lars Hellström