[tex-k] Problem with trip trap test

[Peter Breitenlohner]

On Wed, 3 Aug 2011, Norbert Preining wrote:

> I change also that to generate the correct paths.h (or at least one
> with the right definitions for Debian systems), and build tex, at
> the end the trip trap test breaks.
>
> It seems that the generation of the format (trip.fmt) is not working.
>
> If I try to create trip.fmt according to the triptrap test:
> 	./tex --progname=initex --ini <../../../texk/web2c/triptrap/trip1.in >tripin.fot
> I *do* get a format file (135135 bytes) but the log file tripin.fot
> is a bit strange:
> ---- tripin.fot -----
> This is TeX, Version 3.1415926 (TeX Live 2011/Debian) (INITEX)
> **Please type the name of your input file.
> **(./trip.tex
> ! Bad character code (256).
> <to be read again>
>                   -
> l.26   \nonstopmode\lccode256-
>                              0\mathchardef\a="8000\def\a{ SCALED 3~2769}
> ! Bad mathchar (32768).
> <to be read again>
>                   \def
> l.26 ...opmode\lccode256-0\mathchardef\a="8000\def
>                                                  \a{ SCALED 3~2769}
> ! Illegal magnification has been changed to 1000 (32769).
> .....

Hi Norbert,

this is exactly the expected output, and I also get trip.fmt with 135135
bytes.

> --------------------------
>
> After that, running the next trip test fails with memory allocation. This
> test is in principle:
> 	./tex '&trip trip'
> which produces:
> This is TeX, Version 3.1415926 (TeX Live 2011/Debian)
> fatal: memory exhausted (xmalloc of 18446744071564067984 bytes).
>
> Ouchhhh....

Ouchhhh, indeed.  This number is approx 2^64-2*10^9, so assuming you are on
x86_64, this could be -2*10^9 converted to size_t (which would be `long
unsiged int').  No idea where that negative number would come from.

> Does anyone have an idea where the first problem comes from?

There is no first problem (except that possibly the content of trip.fmt
contains garbage).

Regards
Peter Breitenlohner <peb at mppmu.mpg.de>