Greetings,
sorry for the disturbance; there really was a problem in my code, which
caused spurious SQLEndTran() to be called between the SQLExecute() and the
SQLFetch() -- thus the HY010 was fully justified. What's strange is that
this extra SQLEndTran() doesn't disturb the driver without declare/fetch,
which is why the problem could remain unnoticed for months.
I'd think that an extra mylog("PGAPI_Execute: exit result=%i",retval) just
before the end of PGAPI_Execute() might have helped me find the problem
faster (without creating this list traffic). Also, being able to use
declare/fetch on a selective basis would still be an interesting feature, IMO
(e.g, using declare/fetch unless the statement attrutes force a fallback to
the classic behaviour); unfortunately, I wasn't able to successfully
recompile the driver (Visual Studio 6.0 GUI: yields a corrupted executable
[I'm not a heavy VS user, so I might have done something wrong]; VS 6.0
using NMAKE: a .rc file is missing, killing the build. Cygwin: unhappy at
link time because of an undefined reference to WinMain@16'. What's the
current recommended build environment?)
-- Cyrille
--