IRC log of #zope for Tuesday, 2010-10-19

« Monday, 2010-10-18 Index Wednesday, 2010-10-20 »
donriHelp please: zc.lockfile.LockError: Couldn't lock u'var/db/development.fs.lock'00:03
*** menesis has quit IRC00:05
*** redir has quit IRC00:08
*** tiwula has quit IRC00:16
*** Wu has quit IRC00:16
donriThe IOError is: [Errno 11] Resource temporarily unavailable00:17
*** Wu has joined #zope00:20
CIA-94tseaver * r117736 Zope/src/Zope2/Startup/zopectl.py:00:23
CIA-94Fix ``zopectl start`` on non-Windows platforms.00:23
CIA-94Fixes LP #628448.00:23
CIA-94Forward-port from 2.13 branch.00:23
_mup_Bug #628448: zopectl start - doen't work in Zope2-2.13.0a3 <Zope 2:Fix Committed by tseaver> <https://launchpad.net/bugs/628448>00:23
*** tiwula has joined #zope00:24
donriWorks fine in unit tests but not in the development server :(00:31
*** Hypergraphe has quit IRC00:36
*** Subdino has quit IRC00:41
*** makkalot has quit IRC00:42
*** donri has quit IRC00:53
*** ccomb has quit IRC01:03
*** redir has joined #zope01:05
*** dayne has quit IRC01:09
*** mr_jolly__ has joined #zope01:15
*** hever has joined #zope01:19
*** dvschramm is now known as dvschramm|away01:24
*** mr_jolly has joined #zope01:28
*** J1m has quit IRC01:30
*** MatthewWilkes has quit IRC01:41
*** MrTango has quit IRC01:41
*** mr_jolly has left #zope01:49
*** lucmult has left #zope01:50
*** mr_jolly has joined #zope02:33
*** mr_jolly__ is now known as mr_jolly02:33
*** daMaestro has quit IRC02:57
*** mr_jolly has quit IRC02:59
*** mr_jolly has joined #zope03:12
*** tiwula has quit IRC03:32
*** davisagli is now known as davisagli|away04:05
*** pepeu has joined #zope04:16
*** pepeu has left #zope04:17
*** davisagli|away is now known as davisagli05:21
*** acsr has joined #zope05:26
*** acsr1 has quit IRC05:29
*** hever has quit IRC06:25
*** benji has quit IRC06:54
*** sm has quit IRC07:25
*** kursor has joined #zope07:25
*** menesis has joined #zope08:31
*** davisagli is now known as davisagli|away08:34
*** __mac__ has joined #zope08:43
*** JaRoel|4D has quit IRC08:43
*** alga has joined #zope08:45
*** kursor has quit IRC09:10
*** davisagli|away is now known as davisagli09:15
*** janjaapdriessen has joined #zope09:18
*** Theuni1 has joined #zope09:34
*** alexpilz has joined #zope09:41
*** alexpilz has quit IRC09:46
*** smita has joined #zope09:50
*** janjaapdriessen has quit IRC09:51
CIA-94icemac * r117737 z3c.schema2xml/ (buildout.cfg CHANGES.txt setup.py):09:53
CIA-94* Updated tests to newer Grok version (1.1).09:53
CIA-94* Run tests against lxml 2.2.6.09:53
CIA-94* Added test extra to declare test dependency on ``zope.app.testing``.09:53
CIA-94icemac * r117738 z3c.schema2xml/ (LICENSE.txt bootstrap.py COPYRIGHT.txt): Conform to repository policy.09:53
CIA-94icemac * r117739 z3c.schemadiff/ (4 files in 2 dirs):09:53
CIA-94- Using Python's ``doctest`` module instead of depreacted09:53
CIA-94``zope.testing.doctest``.09:53
CIA-94- Added test extra to declare test dependency on ``zope.component [test]``.09:53
CIA-94icemac * r117740 z3c.schemadiff/ (COPYRIGHT.txt setup.py LICENSE.txt bootstrap.py): Conform to repository policy.09:53
CIA-94icemac * r117741 z3c.searcher/ (CHANGES.txt setup.py src/z3c/searcher/tests.py):09:53
CIA-94- Fixed test dependencies, so tests run with current version of ``z3c.form``.09:53
CIA-94- Using Python's ``doctest`` module instead of depreacted ``zope.testing.doctest``.09:53
*** Deichi has joined #zope09:54
*** neo|4D has joined #zope10:00
*** Theuni1 has quit IRC10:02
*** davisagli is now known as davisagli|away10:03
*** JaRoel|4D has joined #zope10:03
*** digitalmortician has quit IRC10:09
*** goschtl has joined #zope10:17
CIA-94icemac * r117742 z3c.searchfilter/ (COPYRIGHT.txt LICENSE.txt): Conform to repository policy.10:18
CIA-94icemac * r117743 z3c.schema2xml/CHANGES.txt: added version number10:18
*** Theuni1 has joined #zope10:20
*** j-w has joined #zope10:22
*** agroszer has joined #zope10:23
*** digitalmortician has joined #zope10:24
*** planetzopebot has quit IRC10:33
*** planetzopebot has joined #zope10:34
*** JaRoel|4D has quit IRC10:34
*** sunew has joined #zope10:34
*** alexpilz has joined #zope10:34
*** agroszer has quit IRC10:34
*** JaRoel|4D has joined #zope10:35
CIA-94janwijbrand * r117744 zc.catalog/setup.py: move change history near the top of the page for easier access10:41
CIA-94janwijbrand * r117745 zc.catalog/ (CHANGES.txt setup.py): Preparing release 1.510:41
CIA-94janwijbrand * r117746 /zc.catalog/tags/1.5: Tagging 1.510:41
*** hever has joined #zope10:46
*** MrTango has joined #zope10:47
*** agroszer has joined #zope10:50
*** hever has quit IRC10:51
*** kursor has joined #zope10:51
*** d2m has joined #zope11:02
*** ccomb has joined #zope11:03
CIA-94janwijbrand * r117747 zc.catalog/ (CHANGES.txt setup.py): Back to development: 1.611:06
*** vipod has joined #zope11:14
*** janjaapdriessen has joined #zope11:24
*** yvl has quit IRC11:26
*** saju_m has joined #zope11:30
saju_mwhich are the popular python modules used for Social Network Authentication11:31
*** digitalmortician has quit IRC11:32
saju_mi want to login my site using facebook, twitter or gmail account11:33
d2mhttp://oauth.net/code/11:33
*** mr_jolly has joined #zope11:37
*** vipod_ has joined #zope11:39
*** vipod has quit IRC11:41
*** vipod_ is now known as vipod11:41
*** digitalmortician has joined #zope11:41
CIA-94janjaapdriessen * r117748 z3c.flashmessage/ (CHANGES.txt setup.py): Preparing release 1.211:47
CIA-94janjaapdriessen * r117749 /z3c.flashmessage/tags/1.2: Tagging 1.211:47
CIA-94janjaapdriessen * r117750 z3c.flashmessage/ (CHANGES.txt setup.py): Back to development: 1.311:47
CIA-94janjaapdriessen * r117751 groktoolkit/grok.cfg: Pick up z3c.flashmessage, which has less zope.app[testing,zcmlfiles]11:47
*** MatthewWilkes has joined #zope11:52
*** tisto has joined #zope11:55
*** evilbungle has joined #zope12:00
CIA-94janwijbrand 1.2 * r117752 groktoolkit/grok.cfg: add zc.catalog 1.5 that now has only a soft dependency on zope.app.* packages12:03
CIA-94janwijbrand * r117753 groktoolkit/grok.cfg: add zc.catalog 1.5 that now has only a soft dependency on zope.app.* packages12:03
*** Wu has quit IRC12:13
*** Wu has joined #zope12:13
*** Wu has quit IRC12:14
*** Wu has joined #zope12:14
*** cna has joined #zope12:21
*** saju_m has quit IRC12:23
*** MJ has joined #zope12:23
*** menesis has quit IRC12:33
*** milosn has quit IRC12:49
*** CIA-94 has quit IRC12:49
*** Rotonen has quit IRC12:49
*** Rotonen has joined #zope12:50
*** evilbungle has quit IRC12:50
*** milosn has joined #zope12:50
*** menesis has joined #zope12:51
*** CIA-38 has joined #zope12:51
*** teix has joined #zope12:56
*** mr_jolly has left #zope12:58
*** fredvd_ has joined #zope13:03
*** fredvd_ is now known as fredvd13:04
*** yvl has joined #zope13:05
*** tisto is now known as tisto|lunch13:07
*** evilbungle has joined #zope13:15
*** menesis has quit IRC13:26
*** evilbungle has quit IRC13:27
*** evilbungle has joined #zope13:27
*** kleist has joined #zope13:29
*** astoon has joined #zope13:32
*** fredvd|meeting has joined #zope13:32
*** fredvd|meeting has quit IRC13:32
*** lucmult has joined #zope13:33
*** Subdino has joined #zope13:38
*** gwik has joined #zope13:40
*** tisto|lunch is now known as tisto13:59
*** menesis has joined #zope14:10
*** hever has joined #zope14:14
*** cna_ has joined #zope14:22
*** nitrogenycs has joined #zope14:25
*** cna has quit IRC14:25
*** cna_ is now known as cna14:25
*** digitalmortician has quit IRC14:26
*** digitalmortician has joined #zope14:32
*** Subdino has quit IRC14:42
*** vimes656 has joined #zope14:48
*** digitalmortician has quit IRC15:05
*** dayne has joined #zope15:09
*** digitalmortician has joined #zope15:11
*** kursor has quit IRC15:11
*** j-w has quit IRC15:16
*** j-w has joined #zope15:19
*** redir has quit IRC15:20
*** huajie has joined #zope15:25
*** cna_ has joined #zope15:26
*** benji has joined #zope15:28
*** cna_ has quit IRC15:28
*** cna_ has joined #zope15:28
*** cna has quit IRC15:30
*** cna_ is now known as cna15:30
CIA-38adamg * r117754 z3c.recipe.winservice/ (6 files in 2 dirs):15:33
CIA-38New feature: ``runscript`` and ``parameters``.15:33
CIA-38More docs, tests. Some fixes. See CHANGES.txt15:33
CIA-38adamg * r117755 z3c.recipe.winservice/ (CHANGES.txt setup.py): Preparing release 0.7.015:33
CIA-38adamg * r117756 /z3c.recipe.winservice/tags/0.7.0: Tagging 0.7.015:33
CIA-38adamg * r117757 z3c.recipe.winservice/ (CHANGES.txt setup.py): Back to development: 0.7.115:33
*** hever has quit IRC15:57
*** J1m has joined #zope16:03
*** dayne has quit IRC16:07
*** alga has quit IRC16:13
*** dayne has joined #zope16:15
*** digitalmortician has quit IRC16:18
CIA-38janwijbrand * r117758 grokproject/grokproject/ (main.py template/buildout.cfg_tmpl templates.py utils.py): delegate bootstrapping the new project to the bootstrap of that project. by not using zc.buildout in the grokproject command anymore, we prevent version conflicts16:18
*** alga has joined #zope16:18
*** __mac__ has quit IRC16:32
*** __mac__ has joined #zope16:33
CIA-38gotcha * r117759 gocept.selenium/ (20 files in 8 dirs):16:34
CIA-38added tests for Zope 2.12 and Plone 416:34
CIA-38added corresponding buildouts16:34
CIA-38gotcha * r117760 gocept.selenium/ (4 files in 4 dirs): rename class to be more explicit16:34
*** __mac__ has quit IRC16:37
CIA-38gotcha * r117761 gocept.selenium/: update ignore16:49
*** kursor has joined #zope16:49
*** fredvd_ has joined #zope16:54
*** fredvd has quit IRC16:56
*** fredvd_ is now known as fredvd16:56
*** sunew has quit IRC16:58
*** MrWu has joined #zope16:59
*** Theuni1 has quit IRC17:00
*** dvschramm|away is now known as dvschramm17:01
CIA-38janwijbrand * r117762 grokproject/grokproject/template/src/+package+/ (app.txt_tmpl ftesting.zcml_tmpl tests.py_tmpl): use new style test layers in newly created project too17:03
CIA-38gotcha * r117763 gocept.selenium/ (3 files in 3 dirs):17:03
CIA-38tests with plone.app.testing17:03
CIA-38corresponding buildout17:03
*** Subdino has joined #zope17:04
j-wccomb: hi - not sure, but did we plan a meeting?17:14
j-wccomb: if not, I'll mail a next-meeting-proposal17:14
ccombj-w: hi, we had some proposals in the previous mail17:14
ccombquite busy now :(17:15
j-wccomb: np, I'll mail a proposal17:15
ccombmaybe we should move the next meetings in the evening17:15
*** sm has joined #zope17:16
j-wccomb: that would be ok for me17:16
*** pepeu has joined #zope17:17
CIA-38janwijbrand * r117764 grokproject/ (grokproject/template/buildout.cfg_tmpl tests_paste.txt): fix tests. remove tests for --grokversion as we want to change that behaviour anyway17:19
CIA-38gotcha * r117765 /Sandbox/gotcha/z3c.taskqueue/trunk/ (docs/HISTORY.txt setup.py): Preparing release 0.1-alpha-217:19
CIA-38gotcha * r117766 /Sandbox/gotcha/z3c.taskqueue/tags/0.1-alpha-2: Tagging 0.1-alpha-217:19
CIA-38gotcha * r117767 /Sandbox/gotcha/z3c.taskqueue/trunk/ (docs/HISTORY.txt setup.py): Back to development: 0.1-alpha-317:19
CIA-38gotcha * r117768 /Sandbox/gotcha/five.taskqueue/trunk/ (docs/HISTORY.txt setup.py): Preparing release 0.1-alpha-217:19
CIA-38gotcha * r117769 /Sandbox/gotcha/five.taskqueue/tags/0.1-alpha-2: Tagging 0.1-alpha-217:19
CIA-38gotcha * r117770 /Sandbox/gotcha/five.taskqueue/trunk/ (docs/HISTORY.txt setup.py): Back to development: 0.1-alpha-317:19
*** hever has joined #zope17:20
*** fredvd has quit IRC17:20
*** fredvd has joined #zope17:23
*** digitalmortician has joined #zope17:27
*** MrWu has quit IRC17:28
*** MrWu has joined #zope17:28
*** Gogo|tty has quit IRC17:29
CIA-38janjaapdriessen * r117771 grokproject/grokproject/utils.py: Avoid spaces in executable path by using a relative one17:34
*** menesis has quit IRC17:35
*** jim_SFU has joined #zope17:37
*** Theuni1 has joined #zope17:38
*** astoon has quit IRC17:40
*** menesis has joined #zope17:41
*** nitrogenycs has quit IRC17:44
CIA-38janwijbrand * r117772 grokproject/ (CHANGES.txt setup.py): improve pypi page17:48
*** pepeu has quit IRC17:51
*** Gogo|tty has joined #zope17:51
*** TresEquis has joined #zope17:52
*** MJ has quit IRC17:52
*** pepeu has joined #zope17:53
*** cna has quit IRC17:54
Theuni1jo17:59
TresEquisTheuni1, are we about to do meeting, or did I miss it?18:00
Theuni1we're about to18:00
TresEquiscool18:00
Theuni1http://docs.zope.org/zopetoolkit/zope-dev/zope-dev-20101019.html18:01
Theuni1Charlie threw in the request to talk about the bicycle repair kit, but he can't make it.18:01
Theuni1So, who else is around?18:02
*** smita has quit IRC18:02
Theuni1Only the two of us, it seems :)18:03
Theuni1So, last week we did a follow-up on the summit goals. I reminded those that took ownership of some goal in the last summary. And a few of us talked a bit more about the stuff we took onto our plates.18:04
Theuni1TresEquis: is there anything you'd like to talk about or ask about the summit goals?18:04
j-wTheuni1: at least there was an atempt to define functional areas18:07
Theuni1j-w: yeah, thomas started a thread, but unfortunately he got sick right after it. he probably won't be back fully until in two weeks18:08
j-wow, that's not nice, wish him well18:09
TresEquisouch18:09
*** smita has joined #zope18:09
TresEquisTheuni1, I don't really have my head in the game18:09
Theuni1will do18:09
Theuni1TresEquis: ah, k18:09
TresEquisI would like to contribute to the BTK stuff18:09
Theuni1Thought so :)18:09
Theuni1Well, the next bug day needs a date.18:10
Theuni1Lemme check what the doodle says18:10
TresEquisin particular, I plan to focus most of my Zope-related efforts through year's end on my drive to clean up docs / tests for those packages18:10
j-wBTK?18:10
Theuni1It says tuesday or wednesday.18:10
Theuni1Does anyone else have an oppinion on the bug day?18:10
TresEquisj-w, the "bicycle-seat toolkit" subset of ZTK18:11
Theuni1j-w: bicycle (repair|tool) kit18:11
Theuni1bicycle-seat even18:11
j-wah, ok18:11
TresEquiszope.interface, zope.component, etc.18:11
Theuni1TresEquis: you and chris need to work on the mnemonic value of this thing *g18:11
TresEquisI'm probably 80% of the way through zope.interface18:11
j-wthe very core part of the ZTK?18:11
TresEquisTheuni1, I like the "-seat" version because it emphasizes the "take it with you everywhere" bits18:12
TresEquisj-w: right, the most widely-useful bits, even outside the main Zope ecosystem18:12
Theuni1ok, I hereby call next tuesday the october bug day.18:12
TresEquissounds good18:12
Theuni1I still wanted to select some bugs to announce on the list that I'd like to work on - I'll do that later today.18:13
Theuni1So, off to the BTK18:13
Theuni1Too bad Charlie can't be here as he requested the discussion.18:13
Theuni1Something I find interesting is that the BTK seems to have some overlap with the goal to define functional areas.18:13
*** davisagli|away is now known as davisagli18:14
j-wi can see that, yes18:14
TresEquisyup, it's part of a "clumping" strategy18:14
TresEquisit happens to be the clump I use even in non-Zope-related work ;)18:14
*** pepeu has left #zope18:15
Theuni1Sounds like what Thomas described as  "software architecture".18:15
TresEquis"layering" is another term18:15
Theuni1TresEquis: ah, by that you refer to the clumping right?18:15
Theuni1I was referring to the clump that contains zope.interface, ...18:16
TresEquisThat clump is the BTK18:16
*** Deichi has quit IRC18:16
TresEquisI guess you could call it a "software architecture" clump, if you really wanted ;)18:16
Theuni1ACK. Wolfgang and I started to ponder that whole functional area as a part that we (as Zope-the-project") should pay more attention to.18:16
*** Gogo|tty has quit IRC18:17
TresEquisI was thinking that the map showing the "clumps" / functional areas was the "software architectuer"18:17
Theuni1The world did evolve around us and there are some patterns in the wild that seem to be interesting (e.g. ports and adapters) to try getting applications designed nicely.18:17
TresEquis"ports"?18:17
*** MatthewWilkes has quit IRC18:17
Theuni1TresEquis: heh, I noticed that we might already be getting lost in terminology clashes18:17
*** nitrogenycs has joined #zope18:18
Theuni1TresEquis: "ports and adapters" is the name for a pattern AKA hexagonal architecture.18:18
CIA-38ldr * r117773 Products.CMFCore/Products/CMFCore/ (7 files in 3 dirs): Merge in adapterize-wfstatus-wfhistory branch18:18
Theuni1see the c2 wiki18:18
Theuni1That stuff quickly leaps/blurs into the domain modelling community which seems to have some good ideas and then some astronautic stuff.18:18
Theuni1Wolfgang and I experimented with some of the things. The current state of the experiments is here:18:19
Theuni1https://code.gocept.com/hg/public/hex/18:19
*** fredvd is now known as fredvd|shop18:19
Theuni1Anyway, we found that spending some time on the BTK including moving the concepts forward would be good.18:20
Theuni1I would regard the BTK as the stuff you use to help you apply good patterns in your apps without re-implementing them all the time.18:20
* Theuni1 stops blathering.18:20
TresEquisTheuni1:  OK18:20
Theuni1Heh. What was the OK for? :)18:21
* J1m lurks18:21
*** vimes656 has quit IRC18:21
* Theuni1 throws J1m a cookie18:21
* Theuni1 hopes he needs to surface to get the cookie18:22
TresEquisTheuni1:  I don't see "port" spelled out here: https://code.gocept.com/hg/public/hex/file/0da40683781e/doc/terminology.rst18:22
J1mI wish I knew of a sane way to talk about architecture.18:22
*** tiwula has joined #zope18:23
TresEquisor do you mean to refer to http://c2.com/cgi/wiki?PortsAndAdaptersArchitecture18:23
Theuni1TresEquis: i do.18:23
Theuni1our terminology section tries to boil it down a bit. looks like we missed it18:24
Theuni1J1m: whats the insane option?18:24
*** Gogo|tty has joined #zope18:24
* TresEquis doesn't have time to read that this morning ;)18:24
Theuni1TresEquis: sure.18:24
J1mTheuni1, uml/udp18:25
*** daMaestro has joined #zope18:25
Theuni1TresEquis: our current status is: some things and ideas in there help writing clean code a lot by separating different areas of concerns (or domains). however, some domains don't allow strict separation, like persistency, in an obvious fashion.18:25
Theuni1J1m: ack.18:26
J1mI don't think I've seen a good architectural description, except at very high levels.18:26
J1mI haven't seen a process that I felt I could apply.18:27
*** alvaro_o_ has joined #zope18:27
J1mI'm an optimist, so I think a process is feasible and that uml or something like it might play a role.18:27
Theuni1:)18:28
J1mI would like to learn about such a pocess.18:28
J1mprocess.18:28
Theuni1Hmm. In my experience I get better to describe architectures by using natural language and keeping to describe to people who don't know the architecture very well.18:28
Theuni1I think it's not too hard to communicate architecture within a small group of people, but it's hard to write it down for someone to just step in.18:29
J1mActually, it's really hard to capture architecture in a way that you can refer to.18:29
Theuni1I find that understanding architecture also usually requires experience in addition to "just reading" about it. But that may be due to not having had *that* good material yet.18:29
Theuni1J1m: what's architecture anyway?18:30
J1mAns some architectures are too big to express in a conversation or keep in your head, so an oral tradition breaks down.18:30
J1ms/Ans/And18:30
mcdonci think people tend to understand "an architecture" better when an app is built atop it in a domain they understand18:30
mcdonctil then, it's like making them eat vegetables18:31
J1mTo me, an architecture captures design decisions at a higher level than code and at a lower level that system block diagrams.18:31
j-wthe "experience" part is important too: you need have hit your head against a wall before you understand you need something better, architecturally18:31
J1mI would find them useful to reason about and review things like data structures used, and problems like index accumulation and unnecessary indirection, to give some examples.18:32
J1mmcdonc, That's exactly what I want to avoid, easting vegetables, doing sit-ups, which is what's at the bottpm of the UMP slope.18:33
J1mI'm looking for enlightenment. :)18:33
* Theuni1 looks beneath the carpet18:33
mcdoncnone here either18:33
Theuni1J1m: we're working on it ;)18:33
J1ms/UMP/UML18:33
Theuni1J1m: what I notice is that the value is at the point where you need to look at what you have at a higher level.18:34
*** mcdonc has quit IRC18:34
Theuni1So basically you're looking for a representation of the system you have, right?18:34
J1mSpecifically, I'm looking for an exhitectural process/expression that provides enlightenment about design decisions.18:34
*** j-w has quit IRC18:34
Theuni1heh, it's 3:30, everyone drops out ;)18:35
J1mYes, or the system I'm going to build.18:35
* Theuni1 stays a bit18:35
Theuni1J1m: I find patterns to be helpful in that direction. Do you?18:35
J1mNo.18:36
*** MrWu has quit IRC18:36
Theuni1Specific reasons?18:36
J1mA) The "patterns" community seems too wrapped up in defining new jargon that no one understands.18:36
*** MrWu has joined #zope18:36
J1mB) Because patterns is really just capturing decisions.  That is what's important and it tool me too long to realize that a lot of the patterns literature is just blowing smoke.18:37
J1mC) To make recording decisions useful, I think you need some sort of (at the risk of another buzzword) knowledge management system to relate and search problems and solutions.18:38
Theuni1Heh, sounds like you need brains :)18:38
Theuni1(not that you don't have one ...)18:38
J1mD) If we had C, we'd still need better ways to express specific architecture.18:38
Theuni1(I hope that didn't come out wrong)18:38
J1mnp18:39
Theuni1I mean, one indicator for this being a hard problem is that at least at when I studied CS there was basically nothing taught in the realm of architecture that didn't boil down to busy-work.18:40
*** mcdonc has joined #zope18:40
Theuni1So the UML guys come along and try to make the busy-work more valuable by turning UML into code, but that really seems backward.18:40
J1mI've had ideas for C. I even prototyped something using archgenxml at one point. :)  It's a hard problem, but mostly different from architecture.18:40
J1mAgain, I'm looking for enlightenment, not code generation.18:40
Theuni1same here :)18:41
J1mI don't have an answer. I don't even have an idea. I just feel the lack.18:41
Theuni1I agree. :)18:41
*** JaRoel|4D has quit IRC18:42
J1mI didn't mean to turn this into a long discussion. You threw me a cookie. :)18:42
Theuni1I guess we won't solve it today ...18:42
Theuni1I'm happy for the discussion.18:42
Theuni1It's basically what the meeting is for.18:42
J1myup, gotta go. Me too.18:42
Theuni1yay! :)18:43
Theuni1alright, me too.18:43
Theuni1cya18:43
Theuni1(actually i just need to fix some light in the apartment)18:43
*** goschtl has quit IRC18:47
*** benji is now known as benji-lunch18:47
CIA-38janjaapdriessen * r117774 grokcore.startup/ (4 files in 2 dirs): Drop zdaemon support.18:55
*** fredvd|shop is now known as fredvd18:55
planetzopebotFinal Week to Register for Plone Conference 2010 (Plone News)  http://plone.org/events/conferences/bristol-2010/final-week-to-register-for-the-plone-conference19:04
*** giampaolo has quit IRC19:10
CIA-38janjaapdriessen * r117775 zope.testrunner/ (CHANGES.txt setup.py): Preparing release 4.0.019:16
CIA-38janjaapdriessen * r117776 /zope.testrunner/tags/4.0.0: Tagging 4.0.019:16
CIA-38janjaapdriessen * r117777 zope.testrunner/ (CHANGES.txt setup.py): Back to development: 4.1.019:16
CIA-38janjaapdriessen * r117778 zopetoolkit/ztk-versions.cfg: Update zope.testrunner19:16
*** menesis has quit IRC19:16
*** sm_ has joined #zope19:19
*** tisto has quit IRC19:20
*** sm has quit IRC19:22
*** sm_ is now known as sm19:22
*** janjaapdriessen has quit IRC19:25
CIA-38janjaapdriessen * r117779 /Sandbox/janjaapdriessen/buildbot/buildout.cfg: Don't test grokproject on windows19:31
*** lucmult has quit IRC19:33
*** lucmult has joined #zope19:33
*** alga_ has joined #zope19:38
*** alga has quit IRC19:38
*** Subdino has quit IRC19:41
*** Subdino has joined #zope19:42
*** benji-lunch is now known as benji19:42
Subdinohi19:43
*** huajie has quit IRC19:44
SubdinoJ1m: would you have some minutes to discuss about the TID vs Serial thing ? Especially, I would like to understand more what prevents invalidations from being sent "out of order"19:44
Subdino(as chan is quiet, I'll risk myself at "pipelining" some ideas ahead of feedback)19:47
*** smita has quit IRC19:47
Subdinoso, I wonder if TID, aka Serial, really needs to be consistent from its allocation (sometime during 2PC) to its persistent store (tpc_finish & invalidation messages)19:48
Subdinobasically, I don't see a problem with generating a first temporary transaction ID, which does not carry any ordering meaning, and upon tpc_finish a defiintive transaction ID, maybe having an ordering meaning19:49
Subdinoreducing the execution window which must be serialised to have strictly increasing invalidation numbers from the whole 2PC (actually, the time between TID allocation and tpc_finish) to tpc_finish alone19:50
Subdinothis is for the first point19:50
Subdinothen, do invalidation really need to be sent in a strictly increasing manner ? as I understand, the only thing which puts this constraint is their role in transaction isolation19:51
Subdinoie, the first invalidation seen after starting a transaction is the first serial this transaction cannot see19:51
Subdinobut then, this could be defined per-object-and-transaction instead of per transaction only I think19:52
Subdinomy idea would be to have a mapping {oid: serial} which would be looked up when deciding if we should issue a load or a loadBefore19:53
Subdinothis is the second (and most important to me) point19:53
Subdinoplease puch holes in those ideas if they have weaknesses19:53
*** sunew has joined #zope19:53
*** webmaven has joined #zope19:54
*** alexpilz has quit IRC19:57
*** MrWu has quit IRC20:01
*** menesis has joined #zope20:16
*** agroszer_ has joined #zope20:20
*** agroszer has quit IRC20:20
*** giampaolo has joined #zope20:23
*** zagy has joined #zope20:29
*** evilbungle has quit IRC20:37
*** gwik has quit IRC20:44
*** redir has joined #zope20:45
J1mSubdino, wrt to the "critical region" for allocating tids, it is definiately between vote and finish (assuming tids aren't returned by store.21:14
J1mAlso note that application code doesn't see tids until after commit.21:15
J1mwrt relative ordering of tids, it's possible that this could be done on a per-object basis, I'd have to think about it.  I think if you did that though, it would be hard or impossible to reason about whether revisions of separate objects were consistent.21:17
J1mSubdino, ayt?21:17
*** hever has quit IRC21:18
*** mcdonc has quit IRC21:19
*** mcdonc has joined #zope21:23
*** menesis1 has joined #zope21:23
*** menesis has quit IRC21:23
*** sunew has quit IRC21:27
SubdinoJ1m: back, reading21:28
Subdinoabout isolation, I think "consistent" should be precisely defined to know if this scheme passes21:29
J1mI have no idea what that means.21:30
Subdinoin my understanding, this scheme is consistent if: "no transaction can break consistency" and "transaction isolation is granted" (ie, it makes no difference if another transaction is running in parallel or not)21:31
Subdinogah, circular definition of consistency21:32
J1mYes, exactly.21:32
J1mThanks for saying it for me. :)21:32
Subdino:)21:32
J1mYou can have multiple transactions commit in parallel, for some definition of parallel, as long as the result is serializable.21:33
J1mConsistency is defined relative to serialization.21:34
*** zagy has quit IRC21:34
Subdinommh, I'm not so sure21:35
J1mThat's the definition of consistency.21:35
SubdinoIIRC, serialisability is a strong level of transaction isolation21:35
J1mYes, and for the most part, that's what ZODB provides.21:35
J1mWith the compromise wrt read isolation.21:36
Theuni1(which in turn is what most DBs provide anyway)21:37
J1myup21:37
Subdinoto take an example, I'm quite sure I can run two concurrent "update" against the same table with no commit ordering constraints on a mysql innodb table21:38
Subdinoas long at they update distinct sets of rows21:38
Subdinoand by comparing table rows to objects in ZODB, I think the same could be applicable21:40
SubdinoI think I mentionned (by mail) a thesis paper about weak consistency21:40
Subdinohttp://www.pmg.lcs.mit.edu/~adya/pubs/phd.pdf21:41
Subdino"Weak Consistency: A Generalized Theory and Optimistic Implementations for Distributed Transactions"21:41
Subdinothis was months ago21:41
J1mYeah, I started to read that and stopped at some point. I forget why. :)21:41
SubdinoI found at that time an old web page (2003-2005, don't remember) mentionning that ZODB might be providing PL-2+ isolation level21:42
SubdinoI won't say that I read it entirely ;)21:42
Subdinoit has the advantage of giving some simple definitions of consistency in examples (sum of two values must be lower than x, each value being independent "objects")21:43
J1mYou're mysql example produces serializable transactions, ignoring  reads.21:44
Subdinothen I believe the same should be possible in zodb too, with snapshot-ish read isolation21:45
J1mThat's what MVCC does.21:45
J1mI don't see how that could work if each object had it's own set of tids.21:45
Subdinosure, but I would like to poke at the "invalidations must be ordered" part of it, and to me this is tightly related to when definitive tid value should be allocated21:46
Subdinoah, in my idea, it's not a set per object21:46
*** zagy has joined #zope21:47
J1mI suspect that IRC isn't the best place to express your idea.21:47
Subdinoit's more an "invalidation_tid" per object... I think I have a patch somewhere that I could paste for comments21:47
J1mPerhaps IRC is a good place to discuss it, after it has been spelled out *clearly*.21:47
*** mcdonc has quit IRC21:47
Subdinommh, I'm not good at this :s21:50
Subdinohttp://pastebin.com/FBpGE7Ts21:50
*** goschtl has joined #zope21:52
J1mA patch is way to low-level anyway.21:53
J1mI would need to see a high-level desctiption of the approach and be convinced that it is correct.21:54
J1mdescriotion21:55
J1mgaaa21:55
Subdinook, I'll try.21:55
Subdinodon't worry :)21:55
*** hever has joined #zope22:05
*** gwik has joined #zope22:12
*** hever has quit IRC22:13
SubdinoJ1m: would you mind if I send you the first version directly, rather than on zodb-dev ? I'm really not comfortable at formalising my ideas (I guess I didn't studied for long enough to learn this).22:17
*** menesis1 has quit IRC22:17
J1mSubdino, that's fine.22:21
Subdinothanks22:21
*** alexpilz has joined #zope22:24
*** agroszer_ has quit IRC22:36
*** kursor has quit IRC22:36
*** daMaestro|isBack has joined #zope22:38
*** daMaestro has quit IRC22:39
*** daMaestro|isBack is now known as daMaestro22:39
*** goschtl has quit IRC22:49
*** teix has left #zope23:02
SubdinoJ1m: sent23:24
Subdinofeel free to criticise my vocabulary, [il]logical constructions, anything23:26
*** RichardBarrell has joined #zope23:28
*** fredvd has quit IRC23:34
*** alvaro_o_ has quit IRC23:39
*** nitrogenycs has quit IRC23:44
*** redir has quit IRC23:52
« Monday, 2010-10-18 Index Wednesday, 2010-10-20 »

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!