[pdftex] Warnings / Parsing pdftex output
a.sporto+bee at gmail.com
Mon Mar 2 13:06:34 CET 2009
parsing output from pdftex is terrible. Unfortunately, the file-line-error
switch does not apply to warnings. We've been using a slightly modified version
of the convertlatexwarnings script when dealing with conference submissions
and seen its limitations. A problem is for instance that the line of context
after an overfull box can be split up into many lines and is not delimited in
any way. Therefore, it seems to be impossible to determine where the output of
the compiler starts again. We've improved the situation locally by changing our
texmf.cnf file to allow for more chars per line.
So the question is do you guys also see the need for better compiler output?
Unfortunately, I have no idea about web2c. However, I've tried to understand
what's happening in pdftex.web and I believe the output for over-/underfull
boxes is happening in line 20574:
font_in_short_display:=null_font; short_display(list_ptr(r)); print_ln;
The problem seems to be the short_display procedure which doesn't apply any
style of quoting or adds any delimiters to the beginning or end of the output.
Maybe it could be changed to optionally append a char(e.g. ".") to the end of a
line that continues on to the next or maybe the entire output could be made
optional. Or, what would certainly be the best but not simplest option, one
could make it generate warnings in file-line-error style. There are lots of
other places where that would have to be done though I'm afraid...e.g. the
"Non-PDF special ignored" warning.
More information about the pdftex