[tex-live] laulatex quiting: fix your writable cache path

[Taco Hoekwater]

On Jun 26, 2013, at 4:10 PM, Peter Breitenlohner <peb at mppmu.mpg.de> wrote:

> On Wed, 26 Jun 2013, Taco Hoekwater wrote:
> 
>> In that case <assert.h> is apparently the culprit, and the previously
>> proposed patch should be altered to look like this instead:
>> 
>> Index: source/texk/web2c/luatexdir/image/writejpg.w
>> ===================================================================
>> --- source/texk/web2c/luatexdir/image/writejpg.w	(revision 4639)
>> +++ source/texk/web2c/luatexdir/image/writejpg.w	(working copy)
>> @@ -23,8 +23,8 @@
>>    "$Id$"
>>    "$URL$";
>> 
>> +#include "ptexlib.h"
>> #include <assert.h>
>> -#include "ptexlib.h"
>> #include "image/image.h"
>> #include "image/writejpg.h"
> 
> Hi Taco,
> 
> that is one case and may or may not be the culprit, but I really meant ALL
> files.  There is, e.g., font/tt_table.w with
>  #include <stdio.h>
>  #include "ptexlib.h"
> and if that file happens to use off_t/fseeko/ftello it will get the 32bit
> versions.

TTFs cannot be >32bit anyway, so that is probably harmless.

The thing that annoys me about this whole LARGE_FILE stuff (besides the fact that
the linux32 interface is an abomination) is that it should have been tested by the person(s)
that wanted it enabled in the first place. 99.999% of all users can live fine without it.

Anyway, sure I can make that change. Just find all files that include "ptexlib.h" and
make sure that it loads first. Actually, would be possible to add a warning to the
top of ptexlib.h just in case it is not? Is there a #define that can be tested against 
portably? (EOF maybe?)

Best wishes,
Taco