[luatex] string.char

[Taco Hoekwater]

On 04/12/2011 11:05 AM, Heiko Oberdiek wrote:
> On Tue, Apr 05, 2011 at 11:31:50AM +0200, Paul Isambert wrote:
>
>> Le 05/04/2011 10:49, Taco Hoekwater a écrit :
>>> On 04/04/2011 10:01 PM, Paul Isambert wrote:
>>>>> \directlua{tex.settoks(0, "A\string\000B")}
>>>>> has the wrong result "A^^@^^X", thus I conclude that
>>>>> tex.settoks might be the problem.
>>>>
>>>> So it seems you've found yourself a nice bug to report :)
>>>> I say so because I don't think Taco is reading the list much these
>>>> times, so a tracker item would be safer.
>>>
>>> I actually do read the list, but usually have no time to reply these
>>> days.
>>>
>>> In this case, the problem is indeed the settoks as that used
>>> strdup() to get the needed memory (and therefore stops at the
>>> first \000). There was a simple fix, so I have applied something
>>> immediately. If you can, please try rev. 4096.
>>
>> It works very well: tested with tex.settoks(0, string.char(65, 0,
>> 66) and tex.settoks(0, "A\string\000B"), both return "A^^@B".
>
> Thanks Taco and Paul for fixing and testing.

> In package pdftexcmds that emulates \pdfescape... for LuaTeX
> I have now added a workaround for the problematic versions
> (0.41?, ..., 0.65).
>    The endpoint would be revision 4096, but status.luatex_svn
> does not work always as expected:
>
> \catcode`\{=1
> \catcode`\}=2
> \catcode`\#=6
> \def\test#1{%
>    \immediate\write16{%
>      [#1: \directlua{tex.write(status.luatex_#1)}]%
>    }%
> }
> \test{svn}
> \test{version}
> \test{revision}
> \csname @@end\endcsname\end
>
> Result for TL2010:
>    [svn: 3736]
>    [version: 60]
>    [revision: 2]
>
> Result for luatex 0.65.0 from ConTeXt minimals:
>    [svn: -1]
>    [version: 65]
>    [revision: 0]

That is a build problem in the minimals that I have to talk to Mojca 
about. luatex 0.66.0 is the first version that officially fixes the 
problem (to be released after lunch).

Best wishes,
Taco