[Pypgsql-devel] [ pypgsql-Bugs-611979 ] CREATE DATABASE impossible?

[<no...@so...>]

Bugs item #611979, was opened at 2002-09-20 01:11
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=116528&aid=611979&group_id=16528

Category: None
Group: None
Status: Open
Resolution: Postponed
Priority: 2
Submitted By: Nobody/Anonymous (nobody)
Assigned to: Billy G. Allie (ballie01)
Summary: CREATE DATABASE impossible?

Initial Comment:
It appears that the current code-base makes a
create-database call impossible.  Basically, the call
can't happen within a transaction, so the default
operation of wrapping everything in a transaction
prevents the operation from working.  I'm new to the
project, but this would seem to be a minimal workaround:

#378
re_CREATEDB =
re.compile('[\s]*CREATE[\s]+DATABASE[\s]', re.I)

#2518
_badQuery = re_CREATEDB.search(query) or (
	(self.conn.version < 70100) and \
	(re_DRT.search(query) or re_DRI.search(query)
	)
)

However, that doesn't address any other potential calls
that need a non-transactional framework.  Not sure how
you'd want to do this in the general case (probably
another method would be required).

Enjoy, and thanks for the module,
Mike

----------------------------------------------------------------------

>Comment By: Billy G. Allie (ballie01)
Date: 2002-10-20 22:31

Message:
Logged In: YES 
user_id=8500

At least the re_DRT and re_DRI checks.  These are only
needed for  PostgreSQL versions less then 7.1, so they can
definitely go in pyPgSQL 3.0.  They can go in future
versions of 2.x if we decide to drop support for versions of
PostgreSQL erlier than 7.1.  It would sure simplify the
logic of that section of code :-)

----------------------------------------------------------------------

Comment By: Gerhard Häring (ghaering)
Date: 2002-10-20 17:26

Message:
Logged In: YES 
user_id=163326

I'd say keep it as it is for now, mainly because I don't
feel like implementing this and testing it with PostgreSQL
7.0 through 7.3 for the next release.

Billy, your suggestion would be fine with me. Does this mean
we could then remove the existing checks for DDL statements
and just add an according note to the docs?

----------------------------------------------------------------------

Comment By: Billy G. Allie (ballie01)
Date: 2002-10-18 20:22

Message:
Logged In: YES 
user_id=8500

With later versions of PostgreSQL, the number of things that
have to be performed outside of a transaction has decreased.
 I would think that setting autocommit to 1 on the
connection would be an acceptable solution for thoses
operations that have to be outside of a transaction..

Your thoughts on this?

----------------------------------------------------------------------

Comment By: Gerhard Häring (ghaering)
Date: 2002-09-20 01:30

Message:
Logged In: YES 
user_id=163326

I'm not entirely sure what to do about this. You can
(temporarily) set autocommit=1 to work around this:

from pyPgSQL import PgSQL
con = PgSQL.connect()
con.autocommit = 1
cursor = con.cursor()
cursor.execute("create database foo")
con.close()

----------------------------------------------------------------------

Comment By: Mike C. Fletcher (mcfletch)
Date: 2002-09-20 01:12

Message:
Logged In: YES 
user_id=34901

Sorry, this report was from me, just forgot to log in first.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=116528&aid=611979&group_id=16528