"possible bug" report -- not a request for help.
Olaf Weber
> Hi. I don't read this list, so if you want to reply to me please do so
> directly.
...
> Due to the size of the document I am dealing with, and the amount of custom
> style code, I decided to just move LaTeX (2.09) as a whole, along with the
> fonts it used, and run it with the new TeX. What I ran into was --
> -----------------------------------------------
> LaTeX error. See LaTeX manual for explanation.
> Type H <return> for immediate help.
> ! Missing \begin{document}.
> \@latexerr ...for immediate help.}\errmessage {#1}
> -----------------------------------------------
> I finally tracked this down to a line in one of my style documents which was
> supposed to be a comment, and which read as follows:
> ----------------------
> %.se LJDmax = '^M'
> ----------------------
> where the ^M is in fact hex 0D, that is an ASCII CR character. The
> new TeX seems to treat this as a line-end, leading to the error
> since the trailing "'" is not then considered to be a comment.
This is a feature.
The problem is that there are three widely used end-of-line
conventions: LF on UNIX, CRLF on MSDos/Windows and CR on Mac.
The inconvenience of having to convert text files just because the
line endings do not match led people to ask for TeX to recognize all
three kinds (just as PostScript does, for example).
You are correct that this could be documented better.
--
Olaf Weber
Part of being competent is knowing that you might be wrong ... even if
you were correct yesterday with the same answer. -- Barry Kearns
t...@maths.tcd.ie
> > conventions: LF on UNIX, CRLF on MSDos/Windows and CR on Mac.
> TeX is supposed to be sort of frozen, and I had hoped not to run into this sort
> of trouble when I decided to solve my problem the way I did.
But this issue is "system-dependent" and is _not_ frozen.
There is no reason why implementations of TeX
should not accept any EOL character.
Leigh Clayton
> This is a feature.
>
> The problem is that there are three widely used end-of-line
> conventions: LF on UNIX, CRLF on MSDos/Windows and CR on Mac.
Yes, of course I understand this.
> The inconvenience of having to convert text files just because the
> line endings do not match led people to ask for TeX to recognize all
> three kinds (just as PostScript does, for example).
... and this makes sense to me.
> You are correct that this could be documented better.
Which was really my point, though I didn't word my message exactly this way.
TeX is supposed to be sort of frozen, and I had hoped not to run into this sort
of trouble when I decided to solve my problem the way I did. Of course, I
readily concede that the original line of source code was quirky at best, but
then the point was that I'd hoped to be able to avoid messing with that code as
much as possible.
A document somewhere listing known "improvements" since my other release, or
maybe since the "original" Pascal TeX, would have saved me some time (if I'd
been able to find it, of course).
But as I said, I'm OK now, I sent the message in the hope that I might save
someone else from a similar hunt. It seemed little enough to contribute, and
I am very grateful to have access to such a useful and well-integrated set
of programs as TeX/dvips/etc have become these days.
Thanks again .../Leigh
Leigh Clayton
> There is no reason why implementations of TeX
> should not accept any EOL character.
By this logic you could change TeX so that it recognised 'a' as a line end
instead of an alphabetic, and claim that the result wasn't incompatible. If you
did this without notice, though, a lot of existing documents would break. I
don't use a Mac so having TeX recognise Mac files isn't of much use to me
(however useful it probably is to many others). And it *is* a change from the
previous version of TeX that processed my documents as intended by whatever
idiot wrote that line.
But I really don't want to get into a big fight here. I sent my report in the
hope that it might help someone, and as I said I am very grateful for the
excellent software. If the consensus among the developers is that nothing needs
to be done about this issue, then I am perfectly happy to accept that.
Thanks ../Leigh
t...@maths.tcd.ie
> > There is no reason why implementations of TeX
> > should not accept any EOL character.
>
> By this logic you could change TeX so that it recognised 'a' as a line end
> instead of an alphabetic, and claim that the result wasn't incompatible.
The term "frozen" is normally applied to tex.web .
The line-reading module in web2c is all its own,
and so could be changed if desired.
Whether or not that would be a good thing is an entirely different matter.
(Obviously it would not be a good thing to use 'a' as end-of-line character.)
Actually, some implementations of TeX _do_ allow different EOL sequences.