Eric,

I seem to get by the PREPARE just fine but die in COMMIT. I've modified your
code for C and Linux (see support.e attached). The backtrace, however, looks
very much like what my other programs are doing. As you can see, my
printf()s before and after PREPARE are there (no GO TO on SQLERROR) but the
program dies at COMMIT:

(gdb) run
Starting program: /home/thad/optixdev/server6/odbfbd/test/support
[New Thread 16384 (LWP 32043)]
ready to call PREPARE
PREPARE called; ready to call COMMIT

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 32043)]
0x40035fa0 in gds__unregister_cleanup ()
from /opt/firebird/lib/libfbclient.so.1
(gdb) bt
#0 0x40035fa0 in gds__unregister_cleanup ()
from /opt/firebird/lib/libfbclient.so.1
#1 0x40050ff9 in INET_set_clients () from /opt/firebird/lib/libfbclient.so.1
#2 0x4006122f in port::disconnect() () from /opt/firebird/lib/
libfbclient.so.1
#3 0x40055084 in REM_detach_database ()
from /opt/firebird/lib/libfbclient.so.1
#4 0x4004183c in isc_detach_database ()
from /opt/firebird/lib/libfbclient.so.1
#5 0x08049192 in main ()
(gdb)

On Wednesday 03 March 2004 11:39, Eric Boyajian wrote:
> ...
> It seems to me that there are two (related?) problems with the FB 1.5
> implementation of the ESQL API.
>
> 1. The isc_embed_dsql_prepare() function is not working properly, at
> least on Windows. Have you verified with my test program that, on
> Linux, this function is causing the subsequent isc_status() to return
> a value of -901 instead of the expected 0? I don't have a Linux box
> with which to do testing.
>
> 2. The isc_detach_database() causes a segmentation fault on both
> Windows and Linux.
>
> Eric
>...


[Non-text portions of this message have been removed]