I am sorry for kind of reposting this, but I have not got any response to my
patch sent for comment to hackers (Subject: Rollback on error):
http://archives.postgresql.org/pgsql-hackers/2004-09/msg00576.php
I just want to find out, if I should try to solve the issues with this patch
(and add regression tests, docs, etc.) now or leave it till after release of
version 8.0.
Reading responses on my intial post I think the feature is not unwelcome, at
least if implemented well, so that it will not do anything unexpected. I
understand this is beta now, and this is a new feature in psql.
Nevertheless I believe it is good to include the feature now, because
a) It increases the testing of savepoints since more people will use
savepoints (all who activate the mode in psql that my patch provides).
b) Given I finish the open issues, the patch has no backward compatibility
issues with scripts etc., even if you put \set IMPLICIT_SAVEPOINTS 'on' in
.psqlrc. For IMPLICIT_SAVEPOINTS 'off' there is no change at all.
c) The code change is rather local and does not add much complexity.
d) Some people trying 8.0 for the first time might find the current behavior
of psql odd if they are used to oracle, mssql etc. At least at the
interactive level, the patch would give them the option to have their
accustomed way of handling e.g. typos.
Thank you for your time and thank you for any response!
Best Regards,
Michael Paesold
P.S. attached is a version of the patch with more/better comments.
