| 00:00.15 | *** join/#storm wallflower (n=wallflow@ip-205-246-113-216.pool.grokthis.net) |
| 04:58.14 | *** join/#storm lymxz (n=lymxz@123.116.113.99) |
| 07:44.23 | *** join/#storm jukart (n=jukart@85-124-221-45.static.xdsl-line.inode.at) |
| 10:21.35 | *** join/#storm lymxz (n=lymxz@123.116.113.99) |
| 10:21.37 | *** part/#storm lymxz (n=lymxz@123.116.113.99) |
| 10:24.55 | *** join/#storm jukart (n=jukart@85-124-221-45.static.xdsl-line.inode.at) |
| 10:38.50 | *** join/#storm mcella (n=michele@ip-171-30.sn3.eutelia.it) |
| 10:40.32 | *** join/#storm mcella (n=michele@ip-171-30.sn3.eutelia.it) |
| 13:15.42 | *** join/#storm bigdo1 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) |
| 13:26.34 | *** join/#storm LarstiQ (n=larstiq@cust.7.157.adsl.cistron.nl) |
| 15:23.55 | *** join/#storm lymx1 (n=lymxz@123.116.113.99) |
| 15:36.41 | *** part/#storm lymx1 (n=lymxz@123.116.113.99) |
| 16:12.17 | *** join/#storm jukart (n=jukart@85-124-221-45.static.xdsl-line.inode.at) |
| 18:25.15 | *** join/#storm jukart (n=jukart@85-124-221-45.static.xdsl-line.inode.at) |
| 18:42.34 | *** join/#storm oreth| (n=skramred@209.234.229.21) |
| 19:46.52 | oreth| | being new to python... how do i install an Egg file? |
| 19:52.15 | radix | oreth|: easy_install or something? I don't know, I don't really use eggs |
| 19:52.30 | radix | oreth|: I suggest just using setup.py |
| 19:53.03 | oreth| | radix: just trying to install the storm files |
| 19:53.17 | oreth| | however, my computer does like .tar.bz2 |
| 19:53.52 | radix | oreth|: "tar xjf storm-0.11.tar.bz2" |
| 19:54.09 | oreth| | windows :| |
| 19:54.35 | radix | oreth|: ah. |
| 19:54.43 | oreth| | don't have much choice in the matter at wokr |
| 19:54.47 | radix | I guess maybe we should start delivering .zip files. |
| 19:54.54 | oreth| | i know it would work just fine on my home machine |
| 19:56.37 | oreth| | twisted enterprise doesn't seem to like MySql that much |
| 19:56.46 | oreth| | i keep getting disconnected from the db |
| 19:58.35 | radix | did you try cp_reconnect? |
| 19:58.57 | radix | storm basically handles connection lost in the same way that Twisted's ConnectionPool does. |
| 20:00.37 | oreth| | yeah... i've got a cp_reconnect=True in my connection pool |
| 20:00.51 | oreth| | it kinda fluctuates |
| 20:00.59 | radix | oreth|: This is the thing. |
| 20:01.10 | radix | oreth|: When connection to the database is lost, there's really no way to transparently reconnect and continue the transaction. |
| 20:01.31 | radix | oreth|: We *can* reconnect automatically for you, but you have to deal with the fact that your connection was lost and your transaction is dead. |
| 20:01.43 | radix | oreth|: So we automatically reconnect, but still raise an exception. |
| 20:02.00 | oreth| | ok |
| 20:02.10 | radix | The idea is that you should catch that exception and retry, at the application level, whatever you were doing. |
| 20:02.36 | radix | (or just give a "temporary failure" kind of message to your end user, if appropriate) |
| 20:02.45 | oreth| | hrm... |
| 20:02.48 | oreth| | ok |
| 20:03.05 | oreth| | that actually makes alot more sense than what i understood last week :P |
| 20:03.21 | radix | yeah. I remember that conversation in #twisted, but I didn't get the chance to really explain it thoroughly. |
| 20:03.32 | radix | (maybe I read it in backscroll or something) |
| 20:03.52 | oreth| | one of the answers i got last week was "Doesn't matter. handle the exception." |
| 20:03.57 | oreth| | and I was like "wtf does that even mean? |
| 20:04.02 | oreth| | ok |
| 20:04.24 | oreth| | so, i don't need to have a "self.cp.connect()" at the begining of each function that calls the database, right? |
| 20:04.35 | radix | er, right :) |
| 20:04.50 | oreth| | not everytghing you read on google is correct >:( |
| 20:04.54 | radix | sheesh. |
| 20:04.56 | radix | who said that? |
| 20:05.17 | oreth| | EXPERTEXCHANGE or something like that |
| 20:05.18 | oreth| | i dunno :P |
| 20:05.32 | radix | it's unfortunate how many people don't know anything about database transactions. |
| 20:06.30 | radix | (and how many of those people give advice on database programming, and how many of those people write database libraries that they want others to use...) |
| 20:06.59 | oreth| | yeah... |
| 20:07.22 | oreth| | i've done a bit... just a small bit... of database creation and client/server DB transactions |
| 20:07.34 | oreth| | error handling has always been kinda backwards for me |
| 20:07.55 | oreth| | never understood why "auto-retry" was never a part of it. |
| 20:08.47 | radix | well, there are issues with automatically retrying a transaction. certain operations may become invalid in the second try |
| 20:08.57 | oreth| | right |
| 20:09.27 | oreth| | given heavily trafficked databases, things could go severly wrong |
| 20:09.32 | radix | It'd be nice if databases supported transactions that could live beyond a connection, but that's got its own host of problems (how do you time out a transaction, etc) |
| 20:10.26 | oreth| | so... i've got a callback and an error callback |
| 20:10.34 | oreth| | what's the best way to "retry" the transaction? |
| 20:10.48 | oreth| | just have the error callback go to the normal callback? |
| 20:10.51 | oreth| | no wait |
| 20:10.53 | oreth| | that's dumb |
| 20:12.06 | radix | It's application-specific, really. Does it even make sense for you to retry the transaction? you've got to be very careful about non-database interactions which might have already succeeded, and so forth. |
| 20:12.30 | oreth| | they are MOSTLY simple "select" commands |
| 20:12.34 | oreth| | so, in this case, retrying is ok |
| 20:12.39 | radix | ok, yeah. |
| 20:13.02 | oreth| | is there a way to "do this until it doesn't fail" kinda thing? :P |
| 20:13.16 | radix | not any function that I can think of. |
| 20:14.29 | oreth| | ok. |
| 20:14.42 | oreth| | if the function i'm calling is "getRow" for example |
| 20:15.12 | oreth| | self.getRow(arg).addCallback(printRow).addErrback(ErrRow) |
| 20:15.44 | oreth| | def ErrRow(): getRow(arg).addCallback(printRow).addErrback(ErrRow) |
| 20:15.55 | oreth| | with the obviousl mistake that i didn't pass the argument to ErrRow |
| 20:16.00 | oreth| | how.... "bad" is that? |
| 20:16.10 | oreth| | IE: likely hood of stack overflow |
| 20:19.20 | radix | You should only retry if the exception is a connection lost. |
| 20:19.41 | oreth| | that makes perfect sense |
| 20:22.41 | radix | (I.e. adbapi.ConnectionLost) |
| 20:23.41 | oreth| | so |
| 20:23.52 | oreth| | print adbapi.ConnectionLost will return TRUE if the connection is lost |
| 20:24.04 | radix | uh |
| 20:24.07 | radix | no, it's an exception |
| 20:24.13 | radix | it will be raised |
| 20:24.29 | radix | (or, rather, it will be the error result of your deferred) |
| 20:24.40 | oreth| | oh |
| 20:25.29 | oreth| | i just got this: |
| 20:25.35 | oreth| | (incoming paste) |
| 20:25.35 | oreth| | 2007/12/24 14:24 -0600 [LogBot,client] adbapi closing: MySQLdb |
| 20:25.35 | oreth| | 2007/12/24 14:24 -0600 [LogBot,client] Connection lost. |
| 20:25.35 | oreth| | 2007/12/24 14:24 -0600 [-] ][Failure instance: Traceback: <class 'twisted.enterp |
| 20:25.35 | oreth| | rise.adbapi.ConnectionLost'>: |
| 20:27.22 | radix | btw, this conversation should be happening in #twisted |
| 20:27.30 | oreth| | oh |
| 20:27.32 | oreth| | my bad |
| 22:58.45 | *** join/#storm jkakar (n=jkakar@204-174-36-245.unknown-dhcp802.dsl.ucc-net.ca) |