Subject: Re: problems w/ pyPgSQL and the Enterprise HOWTO - msg#00344

Previous Message by Date: click to view message preview

problem with long-running threads

I've been getting the following traceback in my log files... any ideas? I'm using a separate thread for a long-running-query... Traceback (most recent call last): File "/usr/local/lib/python2.2/site-packages/twisted/internet/default.py", line 121, in mainLoop File "/usr/local/lib/python2.2/site-packages/twisted/internet/base.py", line 343, in runUntilCurrent File "/usr/local/lib/python2.2/site-packages/twisted/internet/defer.py", line 162, in callback File "/usr/local/lib/python2.2/site-packages/twisted/internet/defer.py", line 213, in _startRunCallbacks --- <exception caught here> --- File "/usr/local/lib/python2.2/site-packages/twisted/internet/defer.py", line 227, in _runCallbacks File "server.py", line 13, in bad req.finish() File "/usr/local/lib/python2.2/site-packages/twisted/protocols/http.py", line 559, in finish exceptions.AttributeError: Request instance has no attribute 'channel'

Next Message by Date: click to view message preview

Re: SOCKS5 Client if anyone cares

On Mon, 24 Feb 2003 08:46:27 +0200 Dmitry Litovchenko <kwaker@xxxxxxx> wrote: > Sorry no unit tests at this time, I have prepared SOCKS5 unit test at > the very beginning but there were added several parameters so > Protocol constructors don't start correctly. I will fix this > eventually. Basically ditch the regex, we have a command somewhere that does this, don't remember wwhere (isIPAddress I think.) A quick visual scan of the code seems ok - give us some tests and I'll check this in. -- Itamar Shtull-Trauring http://itamarst.org/ http://www.zoteca.com -- Python & Twisted consulting ***> Stop the war: http://unitedforpeace.org/ <***

Previous Message by Thread: click to view message preview

RE: problems w/ pyPgSQL and the Enterprise HOWTO

yep, this really sucks. You cannot do parameterized SQL through pyPgSQL at all. Inside the driver, pyPgSQL just does a string subsitution for the query and submits the whole thing to the database. When you pass parameters to a SQL operation with pyPgSQL is has absolutely no effect on performance, so you might as well just build the entire SQL statement youself and pass it in as a single string. from PgSQL.py: def execute(self, query, *parms): . _qstr = query . self.res = self.conn.conn.query(_qstr % parms) Sean. -----Original Message----- From: twisted-python-admin@xxxxxxxxxxxxxxxxx [mailto:twisted-python-admin@xxxxxxxxxxxxxxxxx]On Behalf Of Andrew Bennetts Sent: Sunday, February 23, 2003 10:18 PM To: twisted-python@xxxxxxxxxxxxxxxxx Subject: Re: [Twisted-Python] problems w/ pyPgSQL and the Enterprise HOWTO On Sun, Feb 23, 2003 at 09:20:23PM -0600, Justin Ryan wrote: > > > The howto also says: > > Also worth noting is that this example assumes that dbmodule uses the > > qmarks paramstyle (see the DB-API specification). > > > > This is the cause of the problem; pyPgSQL uses a different paramstyle. > > > > Maybe this should be made clearer, by inserting a > > # Assumes dbmodule.paramstyle == 'qmarks' > > into the example code, where people will read it. > > Perhaps, However not being familiar with the qmarks paramstyle, I > noticed this bit but most likely skimmed over it because I didn't make > the connection.. Learn something new every day.. > > Is the qmarks paramstyle a required part of the DB API 2.0 spec? No, it's even worse than that: DB API 2.0 doesn't specify which paramstyle to use, it merely states that possible values are "qmark", "numeric", "named", "format", and "pyformat". So a DB API 2.0-compliant module is free to implement any of these it feels like, so long as it sets the correct paramstyle for it. Thus they've defined a spec which allows 5 incompatible flavours. As far as I can see, the only portable way to write queries is to format the queries yourself, which means handling correct quoting yourself. Yuck. > > > Is this a problem with pyPgSQL? something else? pyPgSQL is the > > > _recommended_ module in the enterprise howto, so I didn't expect to run > > > into trouble with it.. > > > > Yeah, that is a good point. The howto should probably use examples that > > work with pyPgSQL, to save confusion. > > > > or at least distinguish. I assumed that the howto was using proper DB > API syntax, and that pyPgSQL is not 100% compliant, but it would be nice > to see the line drawn. I don't know that I've worked with a fully DB > API compliant driver at this point ;p > > Thanks for the clarification... The howto is using proper DB API syntax. The problem is that being DB API 2.0 compliant doesn't help here; qmarks format strings are compliant -- but incompatible with pyformat format strings, which are also compliant. Twisted has nothing to do with this mess, and lets you and your db module sort it out between yourselves :) -Andrew. _______________________________________________ Twisted-Python mailing list Twisted-Python@xxxxxxxxxxxxxxxxx http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python

Next Message by Thread: click to view message preview

Re: problems w/ pyPgSQL and the Enterprise HOWTO

On Mon, Feb 24, 2003 at 03:18:02PM +1100, Andrew Bennetts wrote: > On Sun, Feb 23, 2003 at 09:20:23PM -0600, Justin Ryan wrote: > > > > > [snip] > > > > Is the qmarks paramstyle a required part of the DB API 2.0 spec? > > No, it's even worse than that: DB API 2.0 doesn't specify which paramstyle > to use, it merely states that possible values are "qmark", "numeric", > "named", "format", and "pyformat". So a DB API 2.0-compliant module is free > to implement any of these it feels like, so long as it sets the correct > paramstyle for it. Thus they've defined a spec which allows 5 incompatible > flavours. As far as I can see, the only portable way to write queries is to > format the queries yourself, which means handling correct quoting yourself. > Yuck. > It occurs to me that this might be a useful place to apply localization, gettext style. ;) Now, if only there were an LC_SQL variable... Jp -- "Pascal is Pascal is Pascal is dog meat." -- M. Devine and P. Larson, Computer Science 340 -- up 15 days, 22:29, 4 users, load average: 0.03, 0.01, 0.00 pgpKJzrbXm1tj.pgp Description: PGP signature