[texworks] Problem with line mangling

Stefan Löffler st.loeffler at gmail.com
Thu Jul 7 13:23:46 CEST 2011


Hi,

On 2011-07-07 12:18, Paul A Norman wrote:
> There seem to have been some connected issues for a while possibly as
> far back as 2009 depending on the true status/nature of this report.
> http://code.google.com/p/texworks/issues/detail?id=176&can=1&q=line%20number

I agree with Jonathan here, this was a different problem.

> Definitely pre February 2011
> http://code.google.com/p/texworks/issues/detail?id=469&can=1&q=line%20number

Like Jonathan, I too am weighing the possibility that this is indeed a
Qt issue. Of course, having someone else to blame doesn't help the
normal Tw user much, but maybe it could give us some more insight into
what's going on (and eventually the possibility to report this to Qt
upstream properly).

Anyway, regarding the Qt versions I can give some (approximate) numbers:
Tw on Windows is built using the mingw-cross-env environment. From its
release notes, I compiled the following

Jun 8 2010 - Oct 27 2010: mingw 2.14-2.15, Qt 4.7.0beta1
Oct 27 2010 - Dec 11 2010: mingw 2.16, Qt 4.7.0
Dec 11 2010 - Mar 19 2011: mingw 2.17-2.18, Qt 4.7.1
Mar 19 2011 - Jun 7 2011: mingw 2.19-2.20, Qt 4.7.2

All dates are mingw-cross-env release dates, so add maybe a week or two
until I upgraded and used it for Tw. But it should give some rough
estimates when which version of Qt was in use.
Some additional things:
I don't remember any reports from Ubuntu before the 11.04 release (and
at least one colleague of mine used it to write his thesis). Ubuntu
11.04 (natty) uses Qt 4.7.2, whereas its predecessor, Ubuntu 10.10
(maverick) used Qt 4.7.0.

This doesn't imply any hard facts, of course, but still suggests that
the problem was not present int Qt 4.7.0. Together with the pre Feb 2011
condition, this suggests that the problem was introduced in Qt 4.7.1 (if
this theory holds, of course).

One more supporting fact is that, during debugging, I tried to bisect
the source changes to see when the problem occurred, and IIRC I could
reproduce the issue (using Qt *4.7.2*) back to revisions with some 600
revision number (dating back to Apr-May 2010). So, assuming the problem
didn't exist in the builds back then (could someone check?), this is
even more evidence for the Qt origin. Still, we have to find a way to
work around it for now (preferably without the need to patch Qt, which
doesn't work on OSs that use system libraries, such as Linux)...

> Stefan, your point about memory -- my short document was opened second
> after a much much longer one.

I don't know, these are just wild speculations at this point. But
otherwise it's hard to explain why this would be so unreproducible...

HTH,
Stefan


More information about the texworks mailing list