[XeTeX] Seg fault with Charis

[Ross Moore]

Hi Jim,

On 10/11/2007, at 12:28 PM, Hefferon, James S. wrote:

> I'm having a problem with xetex giving me a segmentation fault, and
> I'd appreciate any advice that someone could give.
>

> If I run a small report, say of 8 pages, it runs fine.  I am using  
> longtable
> so I have to run xelatex twice, and it looks great.
>
> Now I switch to a 1500 page report (there are more than 1000  
> longtables).
> The first pass is great.  I can view it and it looks fine.  But I  
> need a
> second pass.  That turns out not so well.
>
If the 1st pass works but the 2nd fails, then there could be something
awkward being written into the .aux file.

Examine the .aux .
If there is a lot of stuff expanded into lain TeX primitives,
e.g. from an unprotected font-switching macro, then you should
try to do something about this.

My **guess** is that this is related to table-captions...

> Here is the last few lines of
> the run:
>
>   Overfull \hbox (7.68636pt too wide) detected at line 816
>    []\EU1/CharisSIL(0)/m/n/10.95 1337
>
>   Overfull \hbox (7.68636pt too wide) detected at line 820
>    []\EU1/CharisSIL(0)/m/n/10.95 1344
>
>   Overfull \hbox (7.68636pt too wide) detected at line 823
>    []\EU1/CharisSIL(0)/m/n/10.95 1349
>   ) [5] [6]Segmentation fault
>   jim at millstone:/home/sphyg/trunk/bin/latex$
>   Output file removed.
>
> It always fails in the same place.  The log file contains roughly the
> same output.
>
> Running the same file with Gentium and two passes works great.   
> Running the
> same file with Gentium twice, and then Charis gives the above  
> error.  That
> makes me suspect that the sheer size is a problem.  Possible?
>
...and some special feature in Charis that is not present in Gentium.
What this feature may be, I cannot say.
But the problem may well go away, with a cleaner .aux file.

> If anyone has any debugging ideas, I'd be glad to hear them.  I am
> ignorant of where to start, I admit.
>
Take a peek at the .aux  file.
If that looks unreadable in places, then that may indicate a possible
cause.

   -rw-r--r-- 1 jim jim  160986 2007-11-09 20:19 College_1.aux



This is quite large; though hard to estimate what it should be,

since maybe you have lots of long captions.


> Looking at the archives
> of this list, I saw a message a few months back about a seg fault
> and Charis, but it seemed to involve small caps, and I don't see why
> a small document would run but not a large one (both contain small
> caps).  But I could be wrong of course.  I'd be happy to supply
> the large file if that would help.
>
> Thanks,
> Jim Hefferon
>

Hope this helps,

	Ross

------------------------------------------------------------------------
Ross Moore                                         ross at maths.mq.edu.au
Mathematics Department                             office: E7A-419
Macquarie University                               tel: +61 +2 9850 8955
Sydney, Australia  2109                            fax: +61 +2 9850 8114
------------------------------------------------------------------------