Em 06/01/2016 19:08, Mark Rotteveel escreveu: > On 6-1-2016 21:09, Dimitry Sibiryakov wrote: >> 06.01.2016 21:02, Adriano dos Santos Fernandes wrote: >>> The bug CORE-5062 is much more deeper than CHAR_TO_UUID and that remains >>> unfixed. >> >> I'd suggest you to rollback your "fix" and at first hand to try to >> remove check for >> exact UUID string length. >> > > I have one concern though, as I indicated in the ticket the problem > **only** occurs if the UUID column (a char(16) octets) has an index. If > the field doesn't have an index, the error does not occur with the exact > same parameter definition! >
Your assumption is not correct. Problems happens even without an index. > So I don't understand why this should be addressed by a fix like this > (https://github.com/FirebirdSQL/core/commit/c5cd8a87e58884f2dc5477dc960eece974242351). > > As it works without an index and fails with an index, to me it seems > that either some sort of conversion is done or not done or character set > information is stripped away when an index is present. > > Also I see one potential flaw with the fix in the above commit: if the > parameter would be in character set octets, then the padding character > is not a space, but 0x00. > CHAR_TO_UUID expects an ASCII string as parameter. Adriano ------------------------------------------------------------------------------ Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
