Re: ODBC driver further developments?

> -----Original Message-----
> From: Bruce Momjian [mailto:pgman@candle.pha.pa.us]
> Sent: 06 April 2004 15:34
> To: Dave Page
> Cc: md-postgres-donath@netapps.de; pgsql-odbc@postgresql.org
> Subject: Re: [ODBC] ODBC driver further developments?
>
>
> Here are the ODBC patches I have kept in my mailbox:
>
>     ftp://candle.pha.pa.us/pub/postgresql/mypatches/odbc.mbox
>
> Anyone want to review/apply them?

Thanks Bruce - there are basically three patches in there:

1) From  Hartmut Raschick

----
- EINTR safety for send/recv, NOT4WIN
- init vars before using them (most prominently in
  dlg_specific.c::SQLGetPrivateProfileString calls),
- avoid those notorious core dumps when logging on hitting
  "null" data... in result.c
- make 07.03.200 work with the new mutex_locks
- may use -DHAVE_ECO_THREAD_LOCKS, if you use libiodbc (or similar)
  which already shells every driver call in mutex_locks
- using the driver (in conjunction w/libiodbc) on Solaris in a
  threaded environment (thus --enable-pthreads) we also found that
  -DHAVE_LOCALTIME_R is a good idea, also configure does not auto-
  link libsocket which is needed anyway... (mind libnsl when not
  threaded)
----

Unfortunately this patch was submitted as updated source files and I've
been completely unsuccessful in reworking them into a patch that is a)
reviewable, and b) able to be applied to cvs tip. This is probably my
cvs/diff inexperience showing through, but whenever I diff the updated
07.03.0200 source against the original, I end up with a 1/2 meg file
containing basically all of each file affected, before and after. I'd
appreciate it if someone else can have a look and see if it is just me
doing something stupid (other than using Windows!), or if you can supply
a diff against CVS tip Hartmut, that'd be great.

2) From Cyrille Ch?p?lov (sorry) - Fixes for compilation using the
Borland compiler.

Patch applied - thanks.

3) From Ludek Finstrle

Patch 1 adds a preselected scheme name to the list of connection/DSN
options. Personally I think this is unecessary because there is already
an 'extra settings' option in which the user can include statements such
as 'set search_path to' if required. The list of options for psqlodbc is
already huge, and in the long term I think we should be looking to
remove options rather than add them (things like KSQO and GEQO can be
configured by the app using SET, or by the DBA if really required)!

Patch 2 adds WIN1250 charset support, and fixes a bug in pgtypes.c. The
first part has been applied, the second is obsolete.

Thanks to all the contributors - sorry it took so long to get to these.

Regards, Dave.