bug-auctex
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Richard Everson] Re: [Bug-AUCTeX] 2007-12-28; TeX-next-error finds


From: Richard Everson
Subject: Re: [Richard Everson] Re: [Bug-AUCTeX] 2007-12-28; TeX-next-error finds wrong file
Date: Sat, 16 Feb 2008 11:41:34 +0000


On 16 Feb 2008, at 11:14, David Kastrup wrote:

Ralf Angeli <address@hidden> writes:

Forwarding this to the mailing list.  Please, keep the mailing list
copied when you are answering.

The problem with the output of the sample file is the following line:

(1in=72.27pt, 1cm=28.45pt)

This confuses the regexp-based matcher in AUCTeX (and preview-latex)
which thinks the period in the numbers denotes the start of a file
extension and following the extension there should not be any
whitespace. This means the opening paren is not found but the closing one is and this will pop the right file name found before from the stack
and only leave the erroneous Web2C stuff in it.

Perhaps we should loosen the restrictions when searching for opening
parens, but then we might have to do the file name matching
differently.  Does somebody have a good idea how to fix this?

This is all such an unreliable mess.  Perhaps our file name recorder
should never forget any file and just record the name and the purported
nesting.  And when an error occurs, it checks every recorded file name
in some intelligent order, preferring loaded files over non-loaded ones
and ignoring non-existent pseudo-files, and then using the first match
for the error context lines.

Or tell people to configure file-line error messages.

file-line error messages fixes both the geometry example and the beamer example (attached as it wasn't copied to the list). Why not invoke "*tex -file-line-error" from AUCTeX, rather than asking people to track down texmf.cnf (or whatever)?

Richard.


Attachment: modelling.tex
Description: Binary data




reply via email to

[Prev in Thread] Current Thread [Next in Thread]