Re: [2.96.1] problem: latex generates pdf rather than dvi

Reinhard Kotucha <[email protected]> wrote:

>>>>>> "Frank" == Frank Küster <[email protected]> writes:
>
>     >> Yes, but I suppose that most problems come from macro packages.
>     >> They have to be fixed anyway.  As far as the documents are
>     >> cocerned, it is not much more work to fix the bugs than to add
>     >> \usepackage{dvi}.
>
>     > Oh yes, it is. It's even easier to say "dvilatex" or "latex
>     > --dvionly". But for sure "latex
>     > '\input{dvi.sty}\input{broken.tex}' is easier than 
>
> Assume that someone writes a package described by David. 

Good assumption. I bet someone will, no I bet there'll be a couple of
them. 

> I strictly recommend not to distribute such a file.  Just checking
> whether \pdfoutput exists and ignoring its value has been as long as
                                                           ^
.an error..  right?

> pdfTeX exists.  
[...]
>     > making a copy
>     > of the document on a rw filesystem, opening it in an editor, and
>     > searching for _all_ places where packages or the author himself
>     > did things wrong. Even more funny when the document uses some
>     > \input or \include calls.
>
> Packages should not be fixed by users.  Users usually do not have
> their document sources on read-only file sytems.

The problem I see is: A user gets some TeX sources from the internet
(say, program documentation) and wants to produce dvi output from them
(for whatever reason, be it he likes xdvi, the document uses pstricks or
whatever). The document is buggy and produces pdf output.

Of course he can write an E-Mail to the author and complain. But he
wants the document today. Although I agree that packages shouldn't be
fixed by user, now our user is facing the problem that he has to find
the buggy code, no matter whether it's in the document or in a package,
and somehow fix it.

> People who are doing
> tricky things are familiar with tools like grep.  There is no need to
> open each file in an editor.

And what about people who don't want to do tricky things, but just get
this cool document typeset without errors?

> What you describe is a "worst case scenario".  Why should anyone put
> tex sources on a read-only file system? 

On my distribution, a whole lot of packages store TeX or texinfo sources
in /usr/share/doc/$packagename, in order to allow users to produce the
output format they like. I assume some of them are buggy.

> And why should they then
> want to recreate the pdf or dvi files again?

Perhaps the package only contains html files, but I want the docs
printed, or it contains only dvi files, but I'd like to have a pdf.

> If you talk about buggy macro packages, they must be fixed anyway.
> And there had been no official teTeX release in the past which
> contained packages with known bugs.

I doubt that, although I didn't follow past releases closely. Do you
expect the same to happen for teTeX 3.0? Do you think Thomas will take
over, just to give one example, the development responsibility of
revtex, or instead drop the package from teTeX?

Summary:

I think it would make life easier to have a really easy means of saying
"Produce DVI under any circumstances". Instead of waiting until many
silly solutions show up in the wild, we should integrate a "clean"[1]
solution in teTeX.

Regards, Frank

[1] Of course it's not really clean, because clean would be to fix the
bugs 
-- 
Frank Küster, Biozentrum der Univ. Basel
Abt. Biophysikalische Chemie