[freetds] sometimes sql is returning wrong data to application

Michal Seliga michal.seliga at visicom.sk
Tue Dec 8 06:25:25 EST 2009


note that i use protocol version 8.0, not 7.x
to me it seems like some trailing 0 is missing... but i don't know where, i
don't understand code very well yet. i am going to try with cvs head version,
just in case that something new already fixed it

Frediano Ziglio wrote:
> Well... the fact that occurs even in no-optimized code is fine, it means
> that is not a compiler or stack overflow problem. dummyTS is used for
> TEXT as timestamp... however timestamp is stored in another place... not
> into text buffer so how could dummyTS finish in text output buffer?
> Probably this is due to a varchar(max) field which is not supported by
> TDS protocol 7.1 so is returned like a TEXT field... have you other
> varchar/text fields in your result? 
> 
> I'll look deeper...
> 
> freddy77
> 
> Il giorno lun, 07/12/2009 alle 16.24 +0100, Michal Seliga ha scritto:
>> update - it happened with not optimized version of odbc library too :( only this
>> time garbage is not so random, if there is problem then there is text 'dummyTS'
>> appended to the end of result. i don't know where it came from
>>
>> Michal Seliga wrote:
>>> this problem is strange and i am not able to find out what happens
>>>
>>> i am using freetds + iodbc + qt and i am connection to mssql 2005 database
>>>
>>> *sometimes* i get wrong result from long varchar field. usually its good when
>>> query is launched at the beginning and it suddnly stop working when it happens
>>> later, when application is running for several minutes.
>>>
>>> when i compiled libtdsodbc.c without optimalizations (-O0 instead of -O2 in
>>> Makefile in odbc directory) then everything started to behave well and this
>>> didn't happened anymore
>>>
>>> it looks to me that something wasn't initialized properly with optimalizations
>>> enabled, but i am not able to find out what exactly it could be. when library
>>> was compiled with -O2 there were some warnings, maybe they were related, maybe
>>> not, i don't know.
>>>
>>> i tried to do some tracing, but with -O0 there is no problem and with -O2
>>> tracing is not very useful so i learned nothing.... i hope developers will have
>>> more luck
>>>
>>>
>>>
>>> _______________________________________________
>>> FreeTDS mailing list
>>> FreeTDS at lists.ibiblio.org
>>> http://lists.ibiblio.org/mailman/listinfo/freetds
>>>
>> _______________________________________________
>> FreeTDS mailing list
>> FreeTDS at lists.ibiblio.org
>> http://lists.ibiblio.org/mailman/listinfo/freetds
> 
> _______________________________________________
> FreeTDS mailing list
> FreeTDS at lists.ibiblio.org
> http://lists.ibiblio.org/mailman/listinfo/freetds
> 



More information about the FreeTDS mailing list