Re: Failure in decoding filter

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 08:35 AM 2002-02-14, Pierangelo Masarati wrote:
>"Kurt D. Zeilenga" wrote:
>> 
>> At 07:48 AM 2002-02-14, Kurt D. Zeilenga wrote:
>> >At 07:40 AM 2002-02-14, Kurt D. Zeilenga wrote:
>> >>At 07:02 AM 2002-02-14, Pierangelo Masarati wrote:
>> >>>  0000:  a4 16 04 02 63 6e 30 10  80 01 61 82 02 64 6f 82
>> >>>  0010:  01 04 82 01 04 82 01 21  30 00
>> >
>> >My built in parser is goofy, too.
>> 
>> Especially when it hasn't been feed enough cold carbonated
>> caffeine.
>> 
>> My initial analysis was correct...  The substrings set is:
>> 
>> >30 10    80 01 61 82 02 64 6f 82 01 04 82 01 04 82 01 21
>> 
>> That's obviously hosed.
>
>The point is: why is it hosed? And how can it change if debug
>is enabled (only TRACE (==0x1) makes it work correctly)?

Well, the first thing I would suggest is to recompile without
any optimization flags set... (or make sure you have a very
stable version of the compiler installed).

>It looks like it is in ber_* encoding/io stuff, because 
>filter parsing/putting has very very little debug code in.
>I really don't know where to start looking...

Secondly, I'd step through the filter code in the debugger
to see if it really is generating multiple FINALs or not...

Note that I wrote a filter test program (ftest)... that
might help.

Kurt