On 7/24/21 12:23 PM, Arnaud Le-roy wrote:
Hi,
This is the complete gdb logs :
>20 frames in a thread without any debug info... nice bug.
Must say that firebird sometimes can create such traces w/o debug info -
this happens when provider is unloaded (by mistake, certainly) in the
middle of
On 7/26/21 5:40 PM, Arnaud Le-roy wrote:
Also I do not see any thread actively working with database, i.e.
probably one with AV (thread 2) is such thread. And in that case
problematic UDF (as always) is first candidate to raise problems. Can
one confirm that w/o UDF bug does not happen?
> Also I do not see any thread actively working with database, i.e. probably
> one with AV (thread 2) is such thread. And in that case problematic UDF (as
> always) is first candidate to raise problems. Can one confirm that w/o UDF
> bug does not happen?
>
Yes w/o UDF the bug does not
>> char *uuid1maccompr()
>
> I wonder how this UDF is declared in the database…
>
DECLARE EXTERNAL FUNCTION F_UUID1MACCOMPR
RETURNS CSTRING(22) FREE_IT
ENTRY_POINT 'uuid1maccompr' MODULE_NAME 'FreeAdhocUDF’;
I have try w/o FREE_IT same behavior
Firebird-Devel mailing list, web interface
> Read, but telling true did not pay enough attention to it.
> Can you provide full build process of UDF? (compiler command line, etc., may
> be build screen log). May be something wrong with switches…
java@debian:~/crash$ cat uuid_functions.h
char * uuid1maccompr();
java@debian:~/crash$
26.07.2021 16:51, Arnaud Le-roy wrote:
char *uuid1maccompr()
I wonder how this UDF is declared in the database...
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
On 26/07/2021 11:51, Arnaud Le-roy wrote:
>
>> Read, but telling true did not pay enough attention to it.
>> Can you provide full build process of UDF? (compiler command line,
>> etc., may be build screen log). May be something wrong with switches…
>
> *java@debian*:*~/crash*$ cat
Hi!
It's a nightmare to work with getInfo API, where client must guess
result buffer size of things it can not know.
Also, buffers are limited to 64K size - due to old API and internal
code. Looks like protocol does not have this restriction currently.
But it seems to me that getInfo may be
Why not return IXpbBuilder? It already has all needed navigation methods.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel