Thursday, 2010-08-05

*** lbieber has joined #openstack00:01
*** pcrews has joined #openstack00:02
*** jdmaturen has quit IRC00:08
*** ptremblett has quit IRC00:13
*** ptremblett has joined #openstack00:13
*** tjyang has joined #openstack00:23
*** aliguori has quit IRC00:32
*** niemeyer has joined #openstack00:32
*** pcrews has quit IRC00:32
*** mmalgeri1 has quit IRC00:37
*** pvo has quit IRC00:37
*** metoikos has quit IRC00:43
*** niemeyer has quit IRC00:50
*** blpiatt has joined #openstack00:51
*** mrayzenoss has joined #openstack00:51
*** mrayzenoss is now known as mray00:51
*** mray has joined #openstack00:51
*** blpiatt has quit IRC00:51
*** ptremblett has quit IRC01:01
*** sirp1 has quit IRC01:03
*** dendrobates is now known as dendro-afk01:07
*** justinsheehy has quit IRC01:13
*** carolinad28 has quit IRC01:16
*** dendro-afk is now known as dendrobates01:17
*** justinsheehy has joined #openstack01:19
*** gustavomzw has left #openstack01:23
*** rdw has quit IRC01:43
*** pvo has joined #openstack01:51
*** pvo has joined #openstack01:51
*** ChanServ sets mode: +v pvo01:51
vish1ok network refactor code is up02:01
vish1please review/hack it apart02:01
*** lbieber has quit IRC02:04
alekibangonotmyname: if you thought Spidergoat or EATR robot were scary, dont try googling 'brain eating vaccines'  (yesterdays top search trend in google)02:05
*** jamiew has quit IRC02:18
*** sophiap has joined #openstack02:18
*** mray has quit IRC02:23
*** mrayzenoss has joined #openstack02:23
*** mrayzenoss is now known as mray02:23
*** mray has joined #openstack02:23
*** ivan has quit IRC02:27
*** ivan has joined #openstack02:29
*** Glace has quit IRC02:32
*** Glace has joined #openstack02:38
*** lbieber has joined #openstack02:45
*** jdmaturen has joined #openstack02:52
*** silassewell has quit IRC02:53
*** gundlach has quit IRC03:02
*** kashyapc has joined #openstack03:13
*** jdmaturen has quit IRC03:17
*** rdw has joined #openstack03:21
*** sophiap has quit IRC03:27
*** sophiap_ has joined #openstack03:27
*** kashyapc has quit IRC03:31
*** mrayzenoss has joined #openstack03:49
*** mrayzenoss is now known as mray03:49
*** mray has joined #openstack03:49
*** justinsb has quit IRC03:57
*** kashyapc has joined #openstack04:00
*** mray has quit IRC04:00
*** sophiap_ has quit IRC04:07
*** rajijoom has joined #openstack04:17
*** codejunkie has quit IRC04:42
*** spackest has joined #openstack04:49
spackestany hadoop folks here?  just wondering about starting to use swift as an hdfs replacement04:49
*** monyetbiru has joined #openstack04:50
*** jaker74 has quit IRC05:02
*** jaker74 has joined #openstack05:14
*** jdmaturen has joined #openstack05:22
*** hornbeck has quit IRC05:26
*** lbieber has quit IRC05:26
*** jdmaturen has quit IRC05:48
*** pvo has quit IRC05:51
*** rajijoom has quit IRC05:54
*** jakedahn has joined #openstack05:54
*** almaisan-away is now known as al-maisan06:25
*** BartVB has joined #openstack06:28
*** allsystemsarego has joined #openstack06:47
*** alekibango has quit IRC06:57
*** brd_from_italy has joined #openstack07:01
*** pstevensza has joined #openstack07:06
*** dmd17 has quit IRC07:10
*** al-maisan is now known as almaisan-away07:14
*** calavera_ has joined #openstack07:16
*** almaisan-away is now known as al-maisan07:20
*** spackest has quit IRC07:41
*** randybias has joined #openstack07:46
*** stewart has quit IRC07:53
*** rajijoom has joined #openstack07:56
*** stewart has joined #openstack08:07
*** zebziggle has quit IRC08:07
*** adjohn_ has joined #openstack08:10
*** adjohn has quit IRC08:12
*** adjohn_ is now known as adjohn08:12
*** miclorb_ has quit IRC08:25
*** miclorb_ has joined #openstack08:39
*** HyperJohnGraham has joined #openstack08:44
*** metoikos has joined #openstack08:47
*** jakedahn has quit IRC08:48
*** calavera_ has quit IRC08:54
*** calavera_ has joined #openstack08:55
*** rajijoom has quit IRC08:59
*** alekibango has joined #openstack09:02
*** monyetbiru has quit IRC09:10
*** calavera_ is now known as calavera09:10
*** rajijoom has joined #openstack09:16
*** rnewson has joined #openstack09:26
*** cjwilson has quit IRC09:42
*** miclorb_ has quit IRC09:43
*** jtdowney has joined #openstack09:44
*** brd_from_italy has quit IRC09:45
*** brd_from_italy has joined #openstack09:46
*** brd_from_italy has quit IRC09:59
*** brd_from_italy has joined #openstack10:08
*** ptremblett has joined #openstack10:15
*** randybias has quit IRC10:16
*** brd_from_italy has quit IRC10:18
*** brd_from_italy has joined #openstack10:42
*** miclorb has joined #openstack10:43
*** stefano_s has quit IRC10:44
*** jtdowney has quit IRC10:45
*** brd_from_italy has quit IRC10:47
*** brd_from_italy has joined #openstack10:49
*** stewart has quit IRC10:59
*** brd_from_italy has quit IRC11:22
*** ptremblett has quit IRC11:22
*** clayg has quit IRC11:22
*** ctennis has quit IRC11:23
*** jtdowney has joined #openstack11:29
*** ptremblett has joined #openstack11:29
*** clayg has joined #openstack11:31
*** brd_from_italy has joined #openstack11:33
*** ctennis has joined #openstack11:39
*** jtdowney has quit IRC11:42
*** justinsheehy has quit IRC11:43
*** justinsheehy has joined #openstack11:54
*** kashyapc has quit IRC11:58
*** kashyapc has joined #openstack11:59
*** rajijoom has quit IRC12:00
*** rackerhacker is now known as rackerhacker_12:01
*** rackerhacker_ is now known as rackerhacker12:01
*** jonesy_ has joined #openstack12:01
*** carolinad28 has joined #openstack12:06
*** justinsheehy has quit IRC12:06
*** silassewell has joined #openstack12:09
*** dendrobates is now known as dendro-afk12:12
*** miclorb has quit IRC12:15
*** miclorb has joined #openstack12:30
*** miclorb has quit IRC12:34
*** rajijoom has joined #openstack12:36
*** justinsheehy has joined #openstack12:39
*** rajijoom has quit IRC12:40
*** jsmith-away is now known as jsmith12:41
*** rajijoom has joined #openstack12:54
*** codejunkie has joined #openstack12:56
*** jdarcy has joined #openstack12:59
*** rajijoom has quit IRC13:03
*** RobertLJ has joined #openstack13:05
*** aliguori has joined #openstack13:06
*** lbieber has joined #openstack13:08
*** RobertLJ has quit IRC13:09
*** sophiap has joined #openstack13:15
*** abecc has joined #openstack13:16
*** RobertLJ has joined #openstack13:17
*** kashyapc has quit IRC13:18
*** pvo has joined #openstack13:19
*** ChanServ sets mode: +v pvo13:19
*** gundlach has joined #openstack13:19
*** RobertLJ has quit IRC13:19
*** niemeyer has joined #openstack13:24
*** lbieber has quit IRC13:24
*** tonywolf has joined #openstack13:27
*** brd_from_italy has quit IRC13:32
*** adrian_otto has joined #openstack13:32
*** gustavomzw has joined #openstack13:35
*** brd_from_italy has joined #openstack13:37
*** brd_from_italy has joined #openstack13:49
*** Podilarius has joined #openstack13:52
*** miclorb_ has joined #openstack13:52
*** sophiap has quit IRC14:02
*** mrayzenoss has joined #openstack14:12
*** mrayzenoss is now known as mray14:12
*** mray has joined #openstack14:12
*** jk0 has quit IRC14:16
*** jk0 has joined #openstack14:17
*** jk0 has joined #openstack14:17
*** ChanServ sets mode: +v jk014:17
*** mhinze has joined #openstack14:26
*** mhinze has left #openstack14:26
*** Guest42248 has joined #openstack14:28
*** silassewell has quit IRC14:31
*** Podilarius has quit IRC14:33
*** zul has quit IRC14:34
*** zul has joined #openstack14:35
*** sirp1 has joined #openstack14:40
*** sirp1 has quit IRC14:41
*** sirp1 has joined #openstack14:42
*** sophiap has joined #openstack14:43
*** pcrews has joined #openstack14:47
*** lbieber has joined #openstack14:52
*** miclorb_ has quit IRC14:54
*** littleidea has joined #openstack14:55
*** Podilarius has joined #openstack14:57
*** jaypipes has joined #openstack15:00
*** worldreptiles has joined #openstack15:05
*** lbieber has quit IRC15:10
*** pstevensza has quit IRC15:14
*** lbieber has joined #openstack15:16
*** niemeyer has quit IRC15:17
*** lbieber has quit IRC15:20
*** monyetbiru has joined #openstack15:21
*** rajijoom has joined #openstack15:23
*** dabo has joined #openstack15:23
*** silassewell has joined #openstack15:25
*** lbieber has joined #openstack15:25
*** monyetbiru has quit IRC15:32
*** jbryce has joined #openstack15:35
*** stewart has joined #openstack15:37
*** Guest42248 has quit IRC15:38
*** lbieber has quit IRC15:43
*** spackest has joined #openstack15:47
letterjmtaylor: What are your thoughts on pulling the swift code out of the debian branch since you have to merge the swift code in anyway to rebuild packages?15:47
creihtmtaylor: also would it be possible to add a series for the debian branch?15:47
creihtso someone could do bzr branch lp:swift/debian15:48
creihtor something like that15:48
*** spackest has quit IRC15:48
*** jbryce has quit IRC15:48
*** calavera has quit IRC15:53
*** Guest27053 has joined #openstack15:54
*** pvo_ has joined #openstack16:00
*** brd_from_italy has quit IRC16:02
*** spackest has joined #openstack16:05
*** aliguori has quit IRC16:08
*** gnrfan has joined #openstack16:08
*** lbieber has joined #openstack16:10
*** pvo_ has quit IRC16:14
*** pvo_ has joined #openstack16:14
*** pvo_ has quit IRC16:22
*** hornbeck has joined #openstack16:23
mtaylorcreiht: yes to series16:24
*** jdmaturen has joined #openstack16:25
mtaylorletterj: well, I've got the branches set up in the way that the launchpad infrastructure wants things for the future (it's not all fully done yet) - but having the packaging branch actually be revision-wise releated to the source branch is part of their toolchain design16:25
*** jdmaturen has quit IRC16:25
*** jdmaturen has joined #openstack16:26
exltmtaylor: what do you mean by revision-wise related?16:27
mtaylorexlt: as in they share a history, so you can merge between them16:28
*** worldreptiles has left #openstack16:29
exltso a merge of the upstream code into a /debian-dir-only branch ingores the upstream history? (just trying to understand)16:32
jaypipesgah, I wish we could use the term "service" instead of "worker"...16:33
*** aliguori has joined #openstack16:34
mtaylorexlt: huh? well, to do a _merge_ into a debian-dir-only branch, the debian dir only branch would have to have some shared history ....16:35
mtaylorexlt: but the nice part about the combined packaging branch is that you _don't_ need to merge anything to make a standard package, you just grab the branch, bzr bd, and you're done... it's only when you're preparing new versions that any merging has to be done16:36
*** kashyapc has joined #openstack16:37
exltthanks mtaylor - that makes sense16:38
*** spackest has quit IRC16:38
*** rajijoom has quit IRC16:38
mtaylorexlt: it took me several iterations of lifeless striking me on the head before it all clicked in for me :)16:39
*** jdarcy has quit IRC16:39
mtaylorexlt: I suppose that would be more sensible if you knew who lifeless was :)16:39
exltbtw - I don't believe that I will be doing ITPs for swift/nova in debian, if someone might wish to work on those - I may have my plate full with a couple other contirbutions I am considering16:41
mtaylorexlt: oh - I was actually just about to file some... so good :)16:42
mtaylorthat works out nicely16:42
*** spackest has joined #openstack16:44
*** stewart has quit IRC16:47
*** stewart has joined #openstack16:50
*** spackest has quit IRC16:59
*** randybias has joined #openstack17:00
*** al-maisan is now known as almaisan-away17:00
*** matclayton has joined #openstack17:11
*** spackest has joined #openstack17:23
*** spackest has quit IRC17:25
*** JimCurry has joined #openstack17:29
*** littleidea has quit IRC17:37
jaypipesgundlach: what mail client do you use?  Your ML posts appear really messed up on my browser...17:53
*** Guest27053 has quit IRC17:54
*** spackest has joined #openstack17:55
*** JimCurry has quit IRC17:58
*** matclayton has left #openstack18:04
*** JimCurry has joined #openstack18:05
*** pcrews is now known as pcrews_lunch18:14
*** JimCurry has quit IRC18:26
*** kashyapc has quit IRC18:27
*** Garo_ has quit IRC18:38
*** ctennis has quit IRC18:38
*** Garo_ has joined #openstack18:38
*** hornbeck has quit IRC18:46
*** justinsheehy has quit IRC18:47
*** ptremblett has quit IRC18:47
*** ctennis has joined #openstack18:50
*** jamiew has joined #openstack18:50
*** spackest has quit IRC18:51
notmynamefirst stats system code for swift pushed to my local branch.
gundlachjaypipes: gmail (via google apps).19:00
*** mray has quit IRC19:00
gundlachhow do they look?  lots of question marks all over them?19:00
*** mrayzenoss has joined #openstack19:01
*** mrayzenoss is now known as mray19:01
*** mray has joined #openstack19:01
letterjmtaylor: Is there a way to auto-merge the code into debian branch?19:08
mtaylorletterj: I'm not sure what you mean?19:09
creihtmtaylor: Is there a way we can have the debian branch auto merge from trunk on each commit to trunk?19:09
creihtto keep it up to date19:09
mtaylorcreiht: because you want the branch up to date? or because you want debs on each push to trunk?19:10
creihtbecause it would be nice for it to be up to date19:10
mtaylorcreiht: (we can certainly make something happen, I just want to make sure I understand what you want)19:10
creihtso if someone pulls it to build their own debs, it is up todate19:10
mtaylorcreiht: sure - I don't see any reason why not19:10
creihtthat would make letterj a lot happier :)19:11
mtaylorhehe. ok.19:11
mtaylorwell, I definitely want to make letterj happy19:11
letterjmtaylor: creiht was able to translate for me  :-)19:11
jaypipesgundlach: yup.19:11
*** littleidea has joined #openstack19:11
gundlachjaypipes: i noticed that on as well -- do they look weird in email form?19:12
*** brd_from_italy has joined #openstack19:12
jaypipesgundlach: not sure yet :) I didn't know about that ML until yesterday, so waiting to see ;)19:12
*** dendro-afk is now known as dendrobates19:13
gundlachjaypipes: i'd like to think that this is a launchpad bug rather than a gmail bug, so that there's a better chance of it being fixed.19:13
jaypipesgundlach: I use gmail as well19:13
jaypipesgundlach: and I've never had any issues....19:13
gundlachjaypipes: gmail apps?  you're fwding another (nongoogle) email address to a gmail apps domain?19:14
jaypipesgundlach: no, just straight gmail.19:14
gundlachjaypipes: yeah, i'm wondering if the translation b/w the two email systems is screwing things up.  For that matter, it might be rackspace's problem ( fwds to a google apps gmail account)19:15
jaypipesgundlach: ah, I see...and that would be Exchange :)19:15
jaypipesvish1: around?19:16
vish1jaypipes: yeah, just now19:16
gundlachjaypipes: well there's hardly a chance of getting *that* fixed19:17
*** justinsheehy has joined #openstack19:17
jaypipesvish1: so...see my latest comment on the network-refactor merge proposal...19:20
jaypipesvish1: I think the biggest issue we have that revolves around the datastore and integration with AuthManager is that the models are following "the Rails way" of putting logic into the model classes...19:21
jaypipesvish1: if we followed less of an ActiveRecord pattern and more of a DataMapper pattern, these issues (circular imports, etc) would go away...19:22
vish1jaypipes: agreed19:23
vish1that was the idea behind AuthManager19:23
jaypipesvish1: of course, that's a big change ;)19:23
jaypipesvish1: I know :)19:23
jaypipesvish1: I'm wondering how I can help push that effort along?19:23
vish1but we have a lot of people from the rails world19:23
vish1and they did what was natural19:23
vish1do you have an idea for how we can keep the Auth objects in sync with whatever backend?19:24
jaypipesvish1: oh, I understand, and it's not that the pattern is bad, just that it doesn't quite fit well in this application :) at least, IMHO.19:24
vish1i don't really have a huge position either way19:24
vish1but consistency is good19:24
jaypipesvish1: I do, but I don't want those ideas to hold up your network-refactor branch. :) so, I'll approve and we can discuss it19:24
vish1in the meantime? do you have a simple suggestion for this small issue19:25
vish1should i just call it or move it back into auth/ somewhere?19:25
*** dabo has quit IRC19:25
jaypipesvish1: is fine with me :)19:26
vish1k doing it19:26
jaypipesvish1: been on vacay all week...been trying to catch up with the voluminous emails spawned from eday, gundlach, yourself and many others!19:27
edayjaypipes: you're missing all the action!19:27
*** littleidea has quit IRC19:28
jaypipeseday: heh, just reading your thread about components with gundlach and jorge took an hour... ;)19:28
jaypipeseday: and FWIW, I agree with your diagram.19:28
*** jsmith is now known as jsmith-away19:28
vish1jaypipes: yeah there has been a lot of discussion19:29
vish1jaypipes: I've been avoiding the architecture discussion as much as possible so I can actually get coding done19:30
vish1although I've thrown in my 2c here and there19:30
jaypipesvish1: lol, understood.19:30
*** dabo has joined #openstack19:31
*** jsmith-away is now known as jsmith19:31
*** justinsheehy has quit IRC19:32
mtaylorjaypipes: you need to learn to take vacation like I do - staying on your computer and working!19:33
jaypipesmtaylor: yeah, no thx ;)19:33
mtaylorjaypipes: come on - it's good... you can catch strep throat that way!19:34
jaypipesmtaylor: :)19:34
jaypipesmtaylor: or I suppose :( would be more appropriate19:34
mtaylormmmm. antibiotics19:34
mtaylorremember back when getting sick meant taking days off from work? how quaint!19:35
vish1ok guys...I'm out for a while -- sprint planning19:35
vish1gl and happy reviewing/patching19:35
jaypipeseday: btw, just checking out routes. nice package.19:37
*** vish1 is now known as vish1-afk19:37
edayjaypipes: gundlach told me about it :)19:37
jaypipeseday: yeah, it's very nice.19:38
gundlachjaypipes: i've been looking at the Data Mapper pattern online since you mentioned it but i don't grok it yet.  how's it different from an ActiveRecord pattern -- is the point that it abstracts away the specific column and table names?19:38
*** jsmith is now known as jsmith-away19:39
jaypipesgundlach: kinda.  it enables POD objects to be consumed/transformed/stored without the POD class having any knowledge about the underlying data store.  The reason this is essential in Nova is highlighted here: (see vish's comments on import order...)19:40
jaypipesgundlach: POD meaning plain old data...19:40
*** spackest has joined #openstack19:41
edaymtaylor: the hudson pep8 graphs are labeled pylint, fyi. can't find where that's generated19:42
mtayloreday: they label?19:42
mtayloreday: the label?19:42
jaypipesgundlach: SQLAlchemy allows one to use either paradigm (sqlalchemy.ext.declarative is ActiveRecord, sqlalchemy's default is more DataMapper).  In any case, Nova's persistence model needs to be cleaned up to be non-Redis-specific and to not store persistent objects in LDAP :)19:43
jaypipesgundlach: and no, I'm not saying we should use SQLA ;)19:43
mtayloreday: it's because there is a pylint parser for the violations module19:43
jaypipesgundlach: just that whatever abstraction layer we write over the data store should use the mapper pattern and not the AR pattern, at least IMHO.19:43
mtayloreday: and instead of writing a new parser module for Violations called pep8, I just filter the pep8 output to match pylint format19:43
mtaylorand run it through that19:43
edaymtaylor: ok :)19:44
*** justinsheehy has joined #openstack19:45
jaypipesgundlach: Table Data Gateway is also a good idea IMHO:
edaymtaylor: i know that's incredibly minor and probably borderline annoying, just something I noticed :)19:45
mtayloreday: no, totally. I should really add a proper pep8 parser19:45
jaypipesgundlach: as close as we can get to passing around POD objects, I'd be happy :)19:45
mtayloreday: but this: perl -ple 's/: ([WE]\d+)/: [$1]/' was all I needed to do the transform - so it won out19:45
gundlachjaypipes: i read vish's comment (out of context) and didn't quite get how it relates, but that's fine.  thanks for the explanation.19:46
jaypipesgundlach: sure, hope I explained it ok...19:46
*** pcrews_lunch is now known as pcrews19:46
edaymtaylor: there's something just wrong about using perl to help check python style :)19:48
mtayloreday: I know19:48
mtayloreday: the one thing that language is good for is being better at sed than sed is though19:48
*** ptremblett has joined #openstack19:49
gundlachvish1: dumb question #2 -- why do you end up with a circular import problem?  module a can import module b who imports module a, no problem -- as long as b doesn't try to import a specific name from a before a has defined it, and doesn't run any code besides class & fn definitions at import time.19:50
*** holoway has quit IRC19:51
*** cglee has joined #openstack19:51
gundlachjaypipes: is this rephrasing of Data Mapper correct?  Instead of a class that is both entity and logic for talking to the database, you have an entity class which only contains data, and a "manager" class that knows how to put that data into a datastore.  yes?19:51
jaypipesgundlach: you got it.19:51
gundlachjaypipes: ah, thanks.  i think of that as the "Entity" pattern.  Design pattern fail!19:52
jaypipesgundlach: no worries, I may have been screwing up my terms, too ;)19:52
*** holoway has joined #openstack19:52
jaypipesgundlach: and re: the question you posed to Vish, it's because the Vpn needs to access the AuthManager to grab the project, and the AuthManager needs to import /network/, and if the Vpn class is in /network/, things, well, blow up ;)19:54
jaypipesgundlach: if the models (data objects) were all defined in a single file and had no logic (and thus no need to query, for instance, the AuthManager), that problem would not exist.19:55
jonesysomeone (eday?) did a swift install on centos and doc'd it. Anyone have a link to that?19:55
* jonesy seems to remember some special cases w/ cent. 19:55
gundlachjaypipes: i don't buy it yet.  Vpn only accesses the AuthManager once you have instantiated a Vpn, right?  So by the time we ask the Auth Manager to import /network/, Vpn has already been defined, /network/ has already been imported earlier, and the import statement is a no-op.  Right?  Or if I'm wrong, what are we doing that forces code to run at import time?19:58
edayjonesy: wasn't me :)19:58
gundlach[if you're sure this is not solveable I'll shut up, but circular ref problems in python are usually solveable without having to move your code between modules like in other languages]19:59
jaypipesgundlach: the AuthManager is a runtime-imported interface...20:01
jaypipesgundlach: but regardless, I've run into this before, and the mapper and table data gateway patterns have solved the issues.20:02
gundlachhokay, i'll shut up ;)20:02
*** silassewell has quit IRC20:02
jaypipesgundlach: no, I don't want you to shut up :)20:02
jaypipesgundlach: I'll try to put together some code that shows it...20:03
gundlachjaypipes: ok, so if AuthManager is runtime-imported, i think that means that at import time, /network/ has finished importing, and Vpn fully exists.20:03
vish1-afkgundlach: not actually sure why it doesn't work.  Perhaps because of the from/import style that we are using?  modules that attempt to import each other fail20:03
*** vish1-afk is now known as vish1-mostly-afk20:04
gundlachvish1-mostly-afk: does your current code have this problem?  I assumed this was theoretical so far.  I could dive in and have a look-see.20:04
jaypipesgundlach: here is what happens when (in vish1-mostly-afk's network-refactor branch) we put the Vpn class definition into /network/, which imports auth.manager, which itself imports network.model:
gundlachah, yeah, the from...import causes the problem.20:09
jaypipesgundlach: so, do we need to from .. import .. as?20:09
gundlach(i assume.) import nova.auth instead, and refer to nova.auth.manager when you need it.20:09
jaypipesgundlach: lemme try20:09
gundlachno, the prob, i bet, is that manager hasn't been defined by the time the import circular loop gets going20:10
vish1-mostly-afki remember trying that at one point and it didn't help20:10
*** mray has quit IRC20:10
*** jsmith-away is now known as jsmith20:10
gundlachhmm, yeah, manager is a module, isn't it.  well, let me repro this and poke at it -- i'm bothered now :)20:11
*** ptremblett has quit IRC20:12
jaypipesgundlach: hmm, well, changing to import nova.auth.manager in all the places that previously did from nova.auth import manager did the trick.  Much more verbose, but it works. :)20:13
gundlachoh good!20:14
*** ptremblett has joined #openstack20:14
gundlachsee the circular ref part of -- just read it before asking you the Q, which is why the answer was on the tip of my tongue about from..import20:14
jaypipesgundlach: have I mentioned today that I find the way Python imports works very annoying/confusing? :)20:14
jaypipesgundlach: ooh, very nice article. cheers.20:15
jonesyI find __import__ confusing.20:15
gundlachjaypipes: not yet :)  what part?  i love 'em (except for when a module runs code at import time, like paramiko, so that everything goes wrong if you're not careful.)20:15
gundlachjonesy: yeah, me too -- i think they're trying yet again to come up with a simple unified import syntax under the hood.  i just avoid __import__ if i can manage to get my work done without magic.20:16
jaypipesgundlach: I'm just used to C++ ;)20:16
jaypipesgundlach: just user error, that's all!20:16
gundlachjaypipes: heh, i was going to say that i like them so much more than C++'s (because I never took compilers classes and so i never got *really* comfortable with what exactly happens during linking)20:16
gundlachi like that importing just means "make a new object and stick some attributes on it, and if we've done this already somewhere else then just point me to the already-imported object"20:17
*** mrayzenoss has joined #openstack20:17
*** mrayzenoss is now known as mray20:17
*** mray has joined #openstack20:17
*** dendrobates is now known as dendro-afk20:18
jonesyso, what was the suggested procedure for getting through the SAIO on a machine with only one disk? Just create a file and loopback mount the file?20:18
*** jamiew has quit IRC20:18
jaypipesgundlach: I'm putting a note in that merge prop about our discussions...vish can edit later on, but at least we know it works.  I still stand by my statement that I think ActiveRecord makes the model too "fat" and makes abstraction of the data store more complicated...20:19
gundlachjaypipes: cool, can't speak to the AR debate but glad to help20:19
jaypipesgundlach: cheers, and thx for the mentoring!20:20
mtaylorjaypipes: ++20:20
gundlachheh np20:20
mtaylorAR terrible. Mapper much better20:20
gundlachjaypipes: does AR feel "fat" because there is biz logic in it, and because the Entity knows too much about how it is persisted?20:20
gundlachi really like the benefit of an AR pattern in that there is only one class related to Patient: It's "Patient".  There's not a "Patient" and "PatientManager" splitting data from logic.20:21
vish1-mostly-afkthe basic argument that one me over is that flat interfaces are easier to use than nested ones20:21
jaypipesgundlach: yes, and yes :)20:21
gundlachIf we wanted to keep the single class while having pluggable backends, we'd just need to have our model classes not inherit from a "ActiveRecordToASqlDb" class, and instead inherit from a "ActiveRecord" class that used the strategy pattern to plug in its back end.20:22
gundlach(heh, i say "just" -- it's a simple matter of programming...)20:22
jaypipesgundlach: yeah, but a Patient knows how to get better, not how to store itself.  The table data gateway and mapper know how to persist the Patient's state...20:22
gundlachwould that "shape" of code relieve your concerns about being too fat?20:22
gundlachjaypipes: ahh, that's a neat point20:23
*** cglee has quit IRC20:23
*** metoikos has quit IRC20:23
jaypipesgundlach: coming from a DB background, I view data and objects that deal with data quite's just my nature I suppose!20:23
gundlachjaypipes: sure, but that's what OO is all about -- letting data and the code dealing with that data travel together20:23
*** littleidea has joined #openstack20:23
gundlachas opposed to a procedural approach where data is handed to some separate logic20:24
jaypipesgundlach: depends...20:24
*** jamiew has joined #openstack20:24
gundlachnot saying it's better, just that it "fits" in my head more naturally to put them together.  tho your point about separating persistence from domain knowledge is a good one.20:24
*** metoikos has joined #openstack20:25
jaypipesgundlach: yeah, so the persistence manager/service shouldn't need to know anything about a data object (other than the object has some way of expressing its metadata and its instance data fields).  And likewise, the data object shouldn't need to know how to store itself, only how to express its own data fields...20:26
jaypipesgundlach: a happy medium can be reached pretty easily using the existing BasicModel class  with a few tweaks I think...20:27
gundlachjaypipes: i could buy that.  re not knowing how to store itself, how would i go about getting all of a Patient's Visit records?  In the AR pattern (with a strategy or direct mapping to backend) I can say myPatient.visits, and they are fetched.20:28
gundlachwould i need to say PatientManager.getVisitsFor(myPatient)?20:28
jaypipesgundlach: well, that's the thing... a Patient and a PatientVisit are different entities.  You would do PatientVisit.find_by(Patient)20:29
jaypipesgundlach: you could, of course, also have in the Patient class:20:29
larissajaypipes: Error: "property" is not a valid command.20:29
gundlachah! but i thought that was the AR pattern -- find_by will necessarily have to talk to the persistence layer, right?20:29
jaypipesdef visits(self):20:30
* gundlach looks at larissa20:30
jaypipes  return PatientVisit.find_by(self)20:30
jaypipesgundlach: or similar :)20:30
jaypipesgundlach: yes, this is the AR pattern, I thought that's what you were talking about?20:30
gundlachpardon me.  i'm asking how i would go about getting all of a Patient's Visit records in the Data Mapper pattern20:31
*** hornbeck has joined #openstack20:31
jaypipesgundlach: ah, sorry, I misunderstood your question...20:31
gundlachThe AR pattern allows me to say myPatient.visits() as you pointed out20:31
jaypipesgundlach: you don't "search" or "find" with the mapper, you only update the records with mapper.  you use table data gateway for searching.20:32
*** jamiew_ has joined #openstack20:32
gundlachOK, so the code to get myPatient's visits would be something like TDG.for(Visit).find_all_for(myPatient)?20:32
*** jamiew has quit IRC20:34
*** jamiew_ is now known as jamiew20:34
jaypipesgundlach: yup :)20:34
*** ptremblett has quit IRC20:35
jaypipesgundlach: or, more realistically, something like this:
*** jturney_ has joined #openstack20:37
jaypipesgundlach: right now, Nova's BasicModel combines the TDG (as its @classmethod's like lookup) with the ActiveRecord pattern as its instance methods...)20:38
jaypipesgundlach: so plural objects are mixed in with single instance objects via the class's kinda funky.20:38
* gundlach seriously fails at bzr20:40
gundlachmichael@conch:~/rackspace/nova$ bzr branch lp:~vishvananda/nova/network-refactor20:40
gundlachBranched 218 revision(s).20:40
gundlachmichael@conch:~/rackspace/nova/network-refactor$ find . -name BasicModel20:40
gundlachno results.20:40
jaypipesgundlach: ack-grep ++20:40
mtaylorack-grep ++20:40
*** ptremblett has joined #openstack20:40
jaypipesgundlach: see /nova/datastore.py20:40
jaypipesgundlach: :)20:41
gundlachoh blech, i fail at life.  find . -name BasicModel?  what is wrong with me20:41
*** jturney_ has quit IRC20:41
jaypipesgundlach: if you haven't noticed by now, I'm a big fan of not mixing singular and plural semantics in a single class...20:41
* gundlach picked the wrong day to stop drinking smokes20:41
jaypipesgundlach: lol :)20:41
gundlachjaypipes: yeah, based on your comment above about plural vs single, and looking at the codepad code, i got the impression that maybe all TDM buys you over AR is... not mixing single vs plural.  was going to read BasicModel some before commenting more :)20:42
jaypipesgundlach: as always, there's a million ways to do something.. just comes down to preference, really I suppose20:43
gundlachjaypipes: that's good to understand... i had thought you were saying that TDM/Data Mapper would help get around coupling problems when we tried to make pluggable datastores.20:43
jaypipesgundlach: a little, it might, but for me it's a personal preference.  AR can certainly be adapted to use a pluggable data store (in fact that's what's in my base-datastore-driver branch...)20:48
jaypipesgundlach: I've just run into a bunch of problems with the fact that some objects are stored in the datastore, but others are stored in LDAP, and the two don't play nicely with each other :)20:49
*** jonesy_ has quit IRC20:50
*** DogWater has quit IRC20:51
gundlachjaypipes: i see.  i have no experience trying to tie two datastores together, so i'll defer to your experience :)20:51
edayyay, 1 file passing pep8/pylint :)20:53
mtayloreday: congrats!20:55
*** aliguori has quit IRC20:57
edayexample style change branch proposed20:59
gundlacheday: do i understand that you haven't yet put any of your wsgi stuff into todd's code?20:59
edaylets see how many flame wars I can start in a week :)20:59
edaygundlach: no, I was waiting to see what Todd thought of this approach first, since he's been doing all the WSGI work so far :)20:59
_0x44eday: I thought you said you weren't as flaming as jaypipes?21:00
claygmtaylor: umm... are "feature requests" the same thing as Blueprints?21:04
mtaylorclayg: depends on whether you talk to jaypipes  or me21:04
claygwe've been talking about updating "content-type" via a post, and we need to like... write some of this down :)21:04
claygoriginally i just thought about putting it in as a bug21:04
*** jet__ has joined #openstack21:04
mtaylorclayg: jaypipes sees them as feature requests - I see them as a description of implementations...21:05
gundlachclayg: you're not the first to ask...21:05
claygmtaylor: well... I asked you, but I'll ask jaypipes too... heck - we'll make it a survey!21:05
eday_0x44: I'd probably lose a battle against jaypipes, if thats what you mean :)21:05
mtaylorclayg: so in my world there would be a bug saying "content types are wrong" and then there would be a blueprint describing the work needed to implement it21:05
_0x44eday: Sun Tzu would say you're a superior general for declining to fight him...21:06
_0x44s/fight/argue with him/21:06
claygmtaylor: that seems reasonable to me21:06
eday_0x44: I'd pick a fight with Sun Tzu to show him he's wrong21:07
_0x44At least you'd have a purpose for it, I'd probably just do so for the entertainment.21:08
*** jet__ has left #openstack21:09
comstudi about have a really stupid scheduler working21:10
eday_0x44: so, is it safe to say most of your arguments are for entertainment only, without any real purpose behind them? :)21:11
*** jsmith has left #openstack21:13
comstudok, small typo21:15
*** jakedahn has joined #openstack21:16
comstudseems to work now21:17
comstudbut just picks node at random21:17
_0x44eday: Not most, only some.21:22
_0x44eday: But come on, it's Sun Tzu... :)21:22
*** allsystemsarego has quit IRC21:30
*** littleidea has quit IRC21:35
*** dendro-afk is now known as dendrobates21:38
*** mray has quit IRC21:39
spackestdid a blog post about installing swift at rackspace here
spackestrackspace nearly twice as fast as amazon21:43
creihtspackest: I wonder if that has anything to do with us having mirrors of the packages on our network21:45
creihtAt least I think we do :)21:46
*** miclorb has joined #openstack21:47
*** dabo has quit IRC21:50
spackestwell, unless you changed the ubuntu image to point to your repos, don't think it could have helped21:52
uvirtbotNew bug: #614057 in swift "can't change content-type with-out reupload" [Wishlist,Opinion]
*** BartVB has quit IRC21:56
edaymtaylor: so, do you know what group you need to be in to be able to change assignees for a blueprint?21:58
comstudhm, guess this is not quite working21:58
comstudlibvir: QEMU error : Domain not found: no domain with matching name 'i-8l8frekn'21:58
edaymtaylor: driver vs maintainer? comstud is not able to switch a blueprint assignment21:58
edayhm, well I guess it must be admins, since I can and am not i the driver group21:59
*** adrian_otto1 has joined #openstack22:02
*** dendrobates is now known as dendro-afk22:03
*** littleidea has joined #openstack22:03
mtayloreday: hrm.22:03
mtayloreday: is comstud in nova-core?22:04
edaymtaylor: don't think so, just nova22:04
*** adrian_otto has quit IRC22:05
mtayloreday: yeah - nova is the public team - anybody can join, nova-core owns the ability to do things like accept/assign blueprints and approve merge requests22:05
comstudi'm in nova-core22:05
comstudi thought22:05
edaycomstud: nope, just checked22:05
comstudi suck22:05
mtaylorphew. I was worried something had gotten weird22:06
pvomtaylor: gotten?22:06
* mtaylor giggles22:06
*** jtdowney has joined #openstack22:06
comstudwell, anyway22:07
mtayloreday: I keep hearing that we're going to develop better policy toward core vs. non-core team members - I think the powers that be have been a little busy22:07
comstuddoesn't look like i can edit blueprints then either22:07
comstudthe scheduler blueprint links to an etherpad..22:07
comstudi've put comments about my work there22:07
comstudand a yet-to-be-explained error i'm seeing22:08
comstud(and linked my branch)22:09
edaymtaylor: ahh, who needs policies? I suggest we take the RMS approach and not require passwords...22:10
mtayloreday: why don't you start doing that on your machines and then tell me how that works out :)22:10
*** littleidea has quit IRC22:11
edaymtaylor: if only people were decent and trustworthy22:12
*** JimCurry has joined #openstack22:15
gundlacheday: i'm having trouble figuring out how to get started attacking the API code.  if you aren't going to push up your wsgi stuff, how do you recommend i get started?  and, any idea what might be a good 1st bug to grab?22:17
gundlachlike, should i branch todd's branch and then copy in your pastebin code and try to get "GET /servers" working? or what22:18
*** JimCurry has quit IRC22:19
edaygundlach: I was holding off until we can sync with Todd, who knows what he has done that hasn't been pushed22:19
gundlacheday: yeah, but i'm blocking on him, so i'd like to do *something*, even if it turns out he's already solved that bit.22:20
edaygundlach: hence my stab at pylint cleanup branch :)22:20
edaygundlach: well, you could always follow his setup for now and work from his API branch (has more done there)22:22
edaygundlach: and just hope he hasn't started those bits22:22
edaygundlach: I was also thinking of starting a tornado->enventlet/wsgi branch for the ec2 API22:23
*** dendro-afk is now known as dendrobates22:23
gundlachshould i be coding directly on his branch, or branching his branch? (aka what's the bzr flow for this case)22:23
mtaylorgundlach: I'd branch his branch22:23
mtaylorgundlach: you can always merge from each other as needed22:24
gundlachmtaylor: thanks, will do.  and then i give him a merge request when i want to... ok22:24
*** jtdowney has quit IRC22:24
mtaylorgundlach: yeah - or you don't have to be quite so formal about it  ... you can just be like "dude, grab my changes from lp:~gunlach/nova/blah"22:24
edaygundlach: yeah, we can also order patches going into trunk too. usually want to commit there as often as we can so we don't have large reviews :)22:25
gundlachah, good to know.  i wasn't aware that someone could merge something without a merge request, but i guess that's dumb22:25
edaygundlach: yup, just bzr merge lp:~someone/nova/your_branch  into your local branch. you can also collaborate on group branches too, but thats a little more hairy22:26
edaygundlach: ie, lp:~nova/nova/shared-branch22:27
gundlachcool, i will run thru the branch-change-merge with todd's now and see if it works.22:28
gundlachi appreciate the pointers, guys22:28
*** miclorb has quit IRC22:28
gundlachhopefully i can get unstuck and start coding something22:28
*** Podilarius has left #openstack22:28
edaygundlach: are you going to just add features, or rework the WSGI stuff?22:29
*** jakedahn has quit IRC22:33
*** brd_from_italy has quit IRC22:33
*** jakedahn has joined #openstack22:35
*** mrayzenoss has joined #openstack22:36
*** mrayzenoss is now known as mray22:36
*** mray has joined #openstack22:36
*** nettok has joined #openstack22:39
gundlacheday: i dunno -- i may add your wsgi stuff in, or i may just write some wsgi apps per your spec and not tie them to anything yet22:39
edaymtaylor: hmm, does tarmac land approved branches right away, or is there some delay? I marked vish's branch approved this morning but it's still not merged22:41
*** littleidea has joined #openstack22:44
sirp1eday: is the wsgi-eventlet code that you guys are working on going to be a component that is reusable across the system (basically a micro framework)?22:45
*** dendrobates is now known as dendro-afk22:47
*** mray has quit IRC22:49
edaysirp1: I guess it could, although it's really just WSGI and Routes...22:49
*** lbieber has quit IRC22:50
edaysirp1: see, we could probably reuse the server stuff in Swift to manage eventlet too22:51
edaysirp1: also see swift/common/ in Swift22:52
*** jonesy_ has joined #openstack22:52
*** nettok has quit IRC22:54
*** littleidea has quit IRC22:54
*** miclorb_ has joined #openstack22:55
*** pvo has quit IRC23:08
creihtThe run_wsgi() stuff could be quite useful23:09
*** rnewson has quit IRC23:14
redboSomething based on multiprocessing might be simpler.23:15
*** abecc has quit IRC23:15
comstudbug in trunk23:16
comstudnew bug23:16
* comstud files23:17
comstudfixes my scheduler problem23:18
*** lbieber has joined #openstack23:19
comstudcool, enough done today23:32
uvirtbotNew bug: #614090 in nova "instances fail to launch due to nova.virt.images.fetch error" [Undecided,Fix committed]
*** dendro-afk is now known as dendrobates23:37
*** adjohn has quit IRC23:41
*** tonywolf has quit IRC23:44
*** miclorb_ has quit IRC23:45
Glacewowi cannot believe how zuora is expensive23:46
*** holoway has quit IRC23:47
*** holoway has joined #openstack23:49
*** lbieber has quit IRC23:50
*** spackest has quit IRC23:55
*** adrian_otto1 has quit IRC23:56
*** miclorb has joined #openstack23:56
*** dendrobates is now known as dendro-afk23:58

Generated by 2.14.0 by Marius Gedminas - find it at!