On Sat, Mar 12, 2016 at 4:41 AM, Pavel Stehule <pavel.stehule@gmail.com> wrote: > I afraid so you try to look on your use case as global/generic issue. The > PL/SQL, ADA. PL/pgSQL are verbose languages, and too shortcuts does the > languages dirty. In this point we have different opinion. > > I proposed some enhanced PLpgSQL API with a possibility to create a > extension that can enforce your requested behave. The implementation can not > be hard, and it can coverage some special/individual requests well.
I'm not at all interested to discuss any of the proposed changes that have already been proposed, because we have already had lengthy discussions on them, and I doubt neither you nor me have nothing to add.
What we need is more input on proposed changes from other companies who are also heavy users of PL/pgSQL.
Only then can we move forward. It's like Robert is saying, there is a risk for bikeshedding here, we must widen our perspectives and get better understanding for how other heavy users are using PL/pgSQL.
I disagree with this opinion - this is community sw, not commercial. We can do nothing if we don't find a agreement.
Pavel, do you know of any such companies?
I know companies with pretty heavy PostgreSQL load and critical applications, but with only ten thousands lines of PL/pgSQL. They has only one request - stability. I talked with Oleg and with other people - and common requests are
* session (global) variables
* global temp tables
* better checking of embedded SQL
* usual possibility to specify left part of assign statement
But these requests depends on development style - it is related to classical usage of stored procedures - the people are interesting about it, because they does porting from DB2 or Oracle to Postgres.
Probably the biggest company with pretty large code of PL/pgSQL was Skype, but I have not any info about current state.