Thursday, 2009-03-19

*** tpb has joined #melange00:00
*** disismt has quit IRC00:21
*** disismt has joined #melange00:21
*** mithro has joined #melange00:34
*** ChanServ sets mode: +v mithro00:34
*** disismt has quit IRC00:51
*** madrazr has joined #melange00:54
summatusmentisis there a wiki component to melange?00:54
*** lh has quit IRC00:58
*** lh has joined #melange00:58
*** ChanServ sets mode: +o lh00:58
*** lh_ has joined #melange01:12
*** lh has quit IRC01:15
*** lh_ is now known as lh01:15
*** ChanServ sets mode: +o lh01:15
*** lh has quit IRC01:25
*** lh has joined #melange01:25
*** forever has joined #melange01:38
*** forever has quit IRC01:44
*** madrazr has left #melange01:48
*** allen__ has joined #melange01:49
*** allen__ is now known as _allen_01:50
*** osgeo_wolf has joined #melange01:57
*** mithro has quit IRC02:19
*** mithro has joined #melange02:20
*** ChanServ sets mode: +v mithro02:20
*** penyaskito has quit IRC02:44
*** madrazr has joined #melange03:15
*** madrazr has quit IRC03:50
*** penyaskito has joined #melange04:18
*** rama has joined #melange04:59
ramahello any body in there?05:03
_allen_rama: hi05:08
*** tlsa has joined #melange05:08
tlsai have a few comments about the org list page05:09
tlsais this the right place to air them?05:09
ramai'm new to #melange IRC05:10
_allen_rama: welcome05:10
_allen_tlsa, I think it would be better if you comment on google group05:11
ramayeah, i love to , can you tell me the mailing list address05:12
*** r0bby_ has joined #melange05:13
_allen_rama, you can find it on the site :
tpbTitle: soc - Google Code (at
ramathanks, is there any special one for GSoC 2009?05:14
_allen_you can find it on the site too.05:15
*** r0bby has quit IRC05:16
ramaso are u a developer on melange?05:17
_allen_rama, yes05:17
ramahow long you experice with melange?05:19
ramai have visited the page,
tpbTitle: GSoC2009Ideas - soc - Ideas List for GSoC 2009 - Google Code (at
ramahope it hasn't update since 21 hours05:22
*** grwi has quit IRC05:26
*** nerus has joined #melange05:31
*** rama has quit IRC05:33
*** SRabbelier has joined #melange06:05
*** ChanServ sets mode: +o SRabbelier06:05
_allen_SRabbelier: hi, long time no see06:06
SRabbelier_allen_: heya06:08
SRabbelier_allen_: long time no see indeed06:08
_allen_SRabbelier: hehe06:09
_allen_SRabbelier: I am wondering whether I can catch up with you06:10
SRabbelier_allen_: catch up?06:10
_allen_SRabbelier: I mean follow the progress of melange06:10
SRabbelier_allen_: ah, of course you can :)06:11
_allen_SRabbelier: it's a long time I didn't contribute to melange06:11
_allen_SRabbelier: :)06:11
*** nerus has quit IRC06:11
SRabbelier_allen_: we got a lot done :)06:11
_allen_SRabbelier: I know, that's great!!06:12
_allen_SRabbelier: and gsoc 2009 is running on melange06:12
SRabbelier_allen_: yuP!06:14
*** BarryCarlyon has quit IRC06:36
mithrowhats the "Home Page Document"?06:38
SRabbeliermithro: The Melange document you want to use as the home page for your organization06:39
mithroI think we should have called thousand parsec the 'AAA Thousand Parsec project' :)06:39
mithroSRabbelier: but what does "home page" mean/06:40
SRabbeliermithro: like
tpb<> (at
SRabbeliermithro: that's the home page for gsoc200906:40
mithroSo the "Thousand Parsec" home page is -
tpbTitle: Thousand Parsec : News (at
mithroWe also have the profile page right?06:41
tpb<> (at
SRabbeliermithro: no, it refers to the Melange home page06:42
tpb<> (at
*** solydzajs has joined #melange06:42
*** ChanServ sets mode: +o solydzajs06:42
mithroSRabbelier: still confused06:42
mithrowhen is it ever used?06:42
SRabbeliermithro: In the left hand menu for example06:43
SRabbeliermithro: click 'Home'?06:43
mithrowhy do I care about the 'Home' link?06:44
mithrois it shown anywhere on the site to students or the public?06:49
mithroshould I just use the same stuff as my description field?06:50
SRabbeliermithro: You can use it for whatever you want06:51
mithrohrm, the rss feed render doesn't like our feed :(06:51
SRabbelierlh: we don't have details on this?06:51
SRabbeliermithro: we use the default Google blog widget06:51
solydzajsSRabbelier: I'm working on new blog gadget, should be done today06:53
SRabbeliersolydzajs: oh?06:53
solydzajsSRabbelier: yep to fix this whitespace issues too06:54
SRabbeliersolydzajs: what are you using for the new widget?06:54
solydzajsSRabbelier: blog gadget06:54
solydzajsSRabbelier: it's open source06:54
SRabbeliersolydzajs: ah, cool06:55
*** mithro has quit IRC07:27
*** brlcad has left #melange07:29
*** Lennie has joined #melange07:50
*** Lennie is now known as Guest3482307:50
*** ChanServ sets mode: +o Guest3482307:51
*** Guest34823 has quit IRC07:51
*** Lennie3 has joined #melange07:52
*** Lennie5 has joined #melange07:52
Lennie5java applet fail :)07:52
*** ChanServ sets mode: +o Lennie507:53
*** Lennie5 is now known as Lennie07:53
*** mithro has joined #melange07:55
*** ChanServ sets mode: +v mithro07:55
SRabbelierLennie: lol, epic fail :P07:57
*** MatthewWilkes has joined #melange08:22
*** allen_ has joined #melange08:23
*** _allen_ has quit IRC08:23
*** allen_ has quit IRC08:24
*** Lennie has quit IRC08:36
*** allen has joined #melange08:49
*** madrazr has joined #melange09:00
*** solydzajs has quit IRC09:47
*** MatthewWilkes is now known as mwilkes|uni09:51
*** allen has left #melange10:41
*** cwant has joined #melange10:55
cwantI have a mentor request, but don't recognise the "link id" -- does anybody know how to see the person's profile?10:56
*** Zi has joined #melange11:06
Ziwhat's "the next generation GSoC"? =\11:06
*** mwilkes|uni is now known as MatthewWilkes11:09
MatthewWilkesZi: A sentence fragment11:19
MatthewWilkesyou missed out the web app bit11:19
Ziohhhhhh LOL11:21
Zithanks =D11:22
ZimIRC's title line cut me off11:22
*** kangaroo has joined #melange11:25
kangarooIs there a particular reason an organization would appear in the list of accepted organizations, but not appear in the list of orgs in "Apply to become a mentor"?11:25
cwantmaybe you need to set up org profile first?11:26
kangarooits not in the list of orgs without profiles11:26
cwantwhat is org name?11:26
*** jphr has joined #melange11:28
cwantkangaroo: I'm not involved with melange development (just so you know just guessing things). Maybe the admin needs to do his agreement/profile first?11:28
*** Zi has left #melange11:28
kangaroopossibly, this is why I'm asking11:29
kangaroomiguel is away this week and we have some mentors chomping at the bit11:29
kangarooyou'd figure that if he hadn't done the profile, he'd be in the list of orgs who havn't done profile11:29
kangaroosomething seems up11:29
cwantah, well, I'm guessing thats the prob -- I'd recommend sending message to melange mailing list or put in melange tracker11:29
jphrHas anyone on OSX had huge problems using the new GSoC site using melange?11:30
cwantthe way it worked for me is I got admin invite, then afterwards I looked at the list and saw my org there. I can't say with certainty that my org wasn't there before doing the admin profile11:31
cwantkangaroo: I see mono on the list -- you are on page 211:31
kangaroocwant: click "Apply to become a mentor"11:31
kangaroogo to page 211:32
durin42jphr: I've not had any issues, can you be more specific?11:32
jphrdurin42: i'm unable to log on, projects list is automatically redirecting me to log on page when I click through external links, yet no form to sign in to11:33
cwantkangaroo: I see. Looks like only 133 orgs can do that currently. I'm guessing the prob is something that miguel has to do before mentors can come on board11:34
cwantkangaroo: are you back up admin?11:34
kangaroocwant: nah joseph is, I'll ping him11:34
durin42jphr: do you have cookies disabled or something?11:34
kangarooI probably should be :)11:34
cwantyeah, good plan. I think he has to sign in and accept the role11:35
jphrdurin42: i haven't done so explicitly, i'll double check though11:35
cwantkangaroo: yeah. might not be safest to have project leader as primary soc admin (they tend to be really busy)11:36
*** nerus has joined #melange11:59
*** jphr has left #melange12:04
*** Dragonxue has quit IRC12:16
*** Merio has joined #melange12:20
*** bcarlyon|laptop has joined #melange12:25
*** adimania has joined #melange12:48
*** James--Crook has joined #melange12:53
*** Lennie has joined #melange12:57
*** ChanServ sets mode: +o Lennie12:57
James--CrookSRabbelier: ..a thought on Issue 368 that might save some implementation work...  create a host only button that makes a *melange document* from the orgs.  It's then up to the host to add any preamble, postamble and publish it... can refine the process later...?  Just a thought.12:58
James--CrookLennie: Hi.  Congrats on the launch!12:58
LennieJames--Crook, thanks didnt went as smootly as I hoped12:59
Lenniebut atleast no big mistakes :D12:59
Lennieor failures :)12:59
James--CrookI heard it was the smoothest ever...12:59
LennieIt is my first GSoC ;p12:59
Lennieso it is true12:59
James--Crook(I'm not contradicting you - it can still be the smoothest ever and not gone as smoothly as you would like)13:00
Lenniethat is true13:00
LennieIn my heart I'm a perfectionist :)13:00
James--CrookLennie: What's your thoughts on trying to collaborate with moinmoin during this GSoC and get moinmoin into Melange?13:01
Lennieno thoughts atm13:02
LennieI dont even know moinmoin ^_^13:02
Lennieis that bad? :D13:03
Lenniefound their home page13:03
tpb<> (at
*** amit8-88 has joined #melange13:05
LennieJames--Crook, might be nice13:11
Lenniebut I havent thought it true13:11
Lennieand I think Pawel might have more solid opinion on this13:12
James--CrookIt's in python, and they have the filesystem (backend) abstracted at least to some extent....13:12
SRabbeliercwant, kangaroo: you still there?13:13
James--CrookI think Pawel, you, Sverre are way busy with Melange for GSoC...  and GSoC/M is an additional time demand... so how to solve...  that's what I'm thinking.13:13
cwantHI SRabbelier13:13
SRabbeliercwant: If no org admin is created the org won't show up13:14
SRabbeliercwant: In order to complete the profile an admin profile has to be created13:14
cwantSRabbelier: yeah, I figured as much (kangaroo is actually the one with the issue, not me)13:15
SRabbeliercwant: ah, you figured correctly :)13:15
cwantSRabbelier: any easy way to look up a person's profile when you have a link id? e.g., when somebody you may not no asks to be a mentor13:15
SRabbelier for example13:16
tpbTitle: Sign In Required (at
Lenniecwant, in the new version the public name will be shown on the request page13:16
Lenniein the mean time you can use the link SRabbelier gave you13:16
cwantOk, thanks very much guys13:17
LennieJames--Crook, fair concern13:17
Lenniebut it also depends on what kind of proposals we will be getting13:17
cwantI actually ended up using the autocomplete on the mentor invite page to try to figure out who it was13:17
cwant(didn't know them)13:18
SRabbelierJames--Crook: I don't think it'll be that bad13:18
SRabbelierJames--Crook: we're almost done with the hardest work :)13:18
SRabbelierJames--Crook: by the time GSoC starts we should have a little less pressure13:18
James--CrookSRabbelier: Sorting through the proposals and giving feedback to the prospective students was the most intense time for Audacity last year, just as a mentoring org.  You're probably right, but I see us as being very busy between now and then too...13:22
Lenniecwant, it is good that you can find your way around the system cwant :)13:22
cwantLennie: yeah, it's a pretty nice and intuitive system so far13:23
SRabbelierJames--Crook: we'll see how it goes :)13:23
Lenniesome people disagree cwant :D13:23
LennieI'm just glad it works out of the box atm :P13:23
cwantall systems have some growing pains, but it's evolving into something that is very usable13:24
Lennieone more critical phase to go:)13:25
James--CrookSRabbelier: did you see my suggestion (in scrollback) that might make implementing Issue 368 easier?13:26
SRabbelierJames--Crook: it won't be a problem13:26
SRabbelierJames--Crook: there's only 150 orgs13:26
SRabbelierJames--Crook: retrieving them all once and listing them on one page won't be a problem13:26
Lennieanyone here who has experience with the webapp of last year and would like to give their impression on how the new system looks so far?13:27
James--CrookSure, but how about making a document?  Then LH can add header and footer to it too, without any extra programming?13:27
haoyuSRabbelier, but.. how about if I still want the old style org list?13:27
James--CrookOh... I've just had an idea.  Back in a moment...13:27
SRabbelierhaoyu: make one yourself13:27
SRabbelierhaoyu: the new org list still contains all information13:28
SRabbelierhaoyu: also, the main page won't contain the list of all orgs, it makes it look cluttered13:29
SRabbelierthe fact that it did so in previous years was not really by design13:29
haoyuSRabbelier, that's indeed13:29
SRabbelierit was because it was easier to implement at that time13:29
* SRabbelier is off to have dinner13:30
Lenniethat's more helpful then make on yourself :D13:30
Lenniehave a good meal13:30
Lenniewont be as good as yesterday I think P?13:30
SRabbelierhehe, probably not13:31
James--CrookSRabbelier: Back now...13:31
James--Crookso will the new page have just one table then, with all 150?13:31
SRabbelierJames--Crook: yup13:31
SRabbelierJames--Crook: well, two :P13:31
SRabbelierJames--Crook: but the first one will diminish in size untill it's empty13:31
SRabbelieranyway, g2g now!13:31
James--CrookOK.  Then if soemone wants something else, they can cut and paste it13:31
SRabbelierbbs :)13:31
James--Crookbye SRabbelier!13:32
LennieJames in an ideal web app13:33
LennieI would imagine a JS powered mega cloud13:33
Lenniewhich ties orgs to keywords13:33
James--CrookHmm :(  So someone would need to cut and paste into a spreadsheet, merge and sort, and could THEN publish as melange doc.13:33
Lennieand you can navigate through the clouds looking for stuff you like :P13:33
LennieI'm just dreaming :D13:34
James--CrookLennie: ...  and that magically directs the right students ot the right orgs and keeps the psam applicants away...  mmmm...  I like....  when is it going ot be ready?13:34
Lennief00d time here as well13:34
LennieJames--Crook, next year maybe :D13:34
Lennieif we find the time to do it13:34
James--CrookYeah...  GSoC project for a student :-)13:34
LennieOr just something Merio will do in his free time he had some ideas13:35
LennieMerio, I'll look at your patch tonight btw for duplicates13:35
*** Lennie is now known as Lennie|Food13:35
Lennie|Food*strange sentence13:35
Lennie|FoodI'll look at your patch for the duplicate view tonight :)13:36
James--CrookLennie: Eet smakelijk13:36
* haoyu wondering if there's a way to use google to search on the ideas page13:38
haoyucool, there's a way to do so!13:40
MerioLennie|Food: that's ok :) For the tag (or orgs) cloud, if you remember we were talking about that jquery plugin... still thinking it would be a nice idea ;)13:40
James--CrookMerio...  add it on the wiki ideas page?13:41
MerioJames--Crook: I think that even if it's time consuming it's not a task that imply all the GSoC time13:42
James--CrookNot understanding.  It refers to what?13:42
MerioJames--Crook: the tag/org cloud13:43
*** amit8-88 has left #melange13:44
James--CrookBasic tags with the org would be good.  I like what Alex has done, but it would be nicer if it were here on Melange.13:44
James--CrookIs that what you're thinking about?13:44
MerioJames--Crook: ahem.. about what you're referring? :P13:44
James--CrookIt would also be nice to have brief descriptions as well as the full ones for each org ...  e.g.  "Audacity - Audio Editor", "Melange - Open Source Collaboration System" etc...13:45
* scorche thinks that "open source" is a bit redundant ;)13:46
James--CrookTags would be too small for a whole GSoC project, but could be part of a general improvement...13:46
James--Crookschultmc: right you are.13:46
James--Crookscorche: right you are, is what I meant.  (an autocomplete bollix)13:47
* schultmc goes back to lurking13:47
James--CrookMerio: - Alex Pico initiated this.13:51
tpbTitle: gsoc2009's Bookmarks on Delicious (at
*** brlcad has joined #melange13:52
*** brlcad has left #melange13:52
MerioJames--Crook: it's fine, you're right, it should be integrated in Melange13:53
MerioJames--Crook: with some fancier viewing ;)13:53
MerioJames--Crook, like autocomplete search org by tag13:54
MerioJames--Crook: let's say, starting to write a tag, and while typing you've a view of the orgs which have all the matches (with highlighted matches)13:55
MerioJames--Crook: the searchable tags should include also the language a student can contribute with13:56
James--CrookMerio: Mmm..  One thing that is nice about project ideas is that most can be expanded or reduced in scope...  So perhaps I take back that 'tags' is too small for one GSoC; It could be made a major project in itself.  :-)13:57
MerioJames--Crook: which was my first worry when I searched for an org last year GSoC :)13:57
James--CrookMerio: +1 on including language.13:57
MerioJames--Crook: Agree :P13:57
James--CrookLast year as a mentor I was interested in what all the different projects did...  I was looking for possible synergies...  So one sees a name like 'Project Sunrise' what on earth is that?13:58
MerioJames--Crook: with searchable ideas list :P but it's optional, because I think GSoC is more on letting students brainstorm their own ideas13:58
James--CrookLots of clicking to see what the orgs were.13:58
James--Crook(Melange and Audacity are hardly self explanatory names :-) )13:59
MerioJames--Crook: agree13:59
MerioJames--Crook: also with umbrella orgs it's not so easy to find out what they're all about13:59
James--CrookUmbrealla orgs are tricky.13:59
*** dhaun has joined #melange13:59
MerioAudacity it's self explanatory if you know something about sound processing ;)13:59
James--CrookOn idea lists I'm very much in favour of orgs maintaining them themsleves on their own wikis, in their own format.13:59
James--Crookho ho.13:59
MerioYes, but if they were also copied and pasted in Melange they can be searched inside the app easily14:00
James--CrookI think though that Melange will have to gain wiki functionality over time...14:00
James--Crookand then people will start federating Melange.14:01
James--CrookIt's interesting, because there has been talk of 'world domination', but different people have different views of what that is.14:02
MerioIn my ideas there should be also something to help GSoC students remain in their communities14:02
James--CrookPlease explain more14:02
MerioI don't know really14:02
MerioThe first thing that come in my mind14:02
MerioIs a sort of stats that reveals the activity of that GSoC project over time (after GSoC), but then it would mean doing something bad for those who cannot make it because of few spare time14:03 already mantains public the commits of svn accounts14:04
James--CrookSourceforge has all kinds of trackers for number of downloads and number of open issues...14:04
James--CrookAnd ohloh has kudo rank of people (which I do not think is such a good thing for community, but that's another discussion)14:05
MerioA completely new section for showcasing each work over time, a sort of "live blog"14:05
James--CrookThat's nice.14:05
James--CrookOne thing I'm thinking...14:05
James--Crookis that projects that produce stable code usually do so by having a modular system.14:05
James--Crookso there can be user contributed plug ins.14:06
*** penyaskito has quit IRC14:06
James--CrookThere isn't yet a good way of managing those sub projects.  In some ways it comes back to umbrella orgs again...14:06
James--CrookSo a concrete example...  Soemone writes a DSP plug in for audacity...14:07
James--Crookas a user how do I find out that it exists, whether it is any good, whether it is still maintained?14:07
James--CrookAnd modules give more kudos to the individuals, there is more sense of ownership.14:07
James--Crook(kudos in normal sense, not kudo rank)14:08
James--CrookI think also people are thinking about melange in slightly the wrong way...14:09
MerioWhat do you mean?14:09
James--CrookIt's as if before wiki was invented, someone said I am going to develop a system that can do to do lists, and documentation pages, and discussions....14:10
James--Crookand then they went off and wrote a separate .py for each function.14:10
James--CrookWard's original wiki is a few hundred lines of perl.  Very understandable at one sitting.  All in one file.14:11
James--CrookIn many ways Melange is 'Wiki with tables'.  Or more precisely...14:11
James--CrookStandard Wiki is a quiz with one question.14:12
James--CrookMelange is a wiki where you can edit the edit page and have more than one field...14:12
James--Crook(is that making sense?)14:12
James--CrookStandard wiki edit page is a quiz with one question.  There is one field, a document field.14:13
Merio(I'm following :))14:14
James--CrookIn melange we could create a quiz with one field, and shazzam, all the pages based on it form a wiki.14:14
James--CrookNow we create a quiz with more than one field, and we get a family of related pages off it.14:14
James--CrookWorking at this meta level looks very powerful.14:15
MerioWhat do you think it could become?14:16
James--CrookYou don't need to go off and implement lots of special cases, anymore than you need to implement a todo list, a discussion page, a module.14:16
James--CrookWell, first thing would be to use it for managing a project.  In Audacity we use Media wiki for lots of info, we have feature requests on it cos it's easier than a tracker for that.14:17
James--CrookNow interestingly GHoP is using a tracker as its management app.14:18
James--CrookSo you get an issue tracker, without writing one.14:19
*** bcarlyon|laptop is now known as BarryCarlyon14:19
James--CrookAnd its integrated withe your more amorphous documentation.14:19
James--CrookIf you can gather information across wiki pages, you cut down the amount of programming you have to do.14:20
James--CrookIs this answering your question, 'what could it become'?14:21
MerioI think yes, it's letting me thinking about it a little bit more14:22
James--CrookOK, so at the moment we think of the only editors for wiki as being people.14:24
James--CrookBut actually we could add bots of various kinds too.14:25
SRabbelierJames--Crook: that doesn't work though, there's too many complex ACL/logic things going on to just have everything work like a wiki14:25
James--CrookSRabbelier: I've seen that issue in another setting.  And I can see the difficulty it creates.14:26
*** adimania has quit IRC14:26
James--CrookSome of the rules are complex rules indeed.  If such and such, and time > so and so, and....14:26
James--CrookThe danger is having this custom ACL stuff on each page that needs it.  It's OK on a small project, but I've seen it become a nightmare on a project with thousands of 'models'.14:27
James--CrookWhat happened there was that the ACL logic got cut and pasted to a new related page with a slightly different view.14:28
SRabbelierJames--Crook: that's why we try to keep our ACL system as simple as possible14:28
SRabbelierJames--Crook: for example14:28
James--CrookSo, is it yet as simple as unix permissions?14:28
tpb<$N> (at
SRabbelierJames--Crook: see that dict ;)14:28
James--Crook..having a look...14:28
SRabbelierJames--Crook: that's the only change required to make the /show page of an org app public to anyone, instead of to only to signed in users14:29
SRabbeliers/of to/of/14:29
SRabbelierJames--Crook: no copy and pasting on my watch ;)14:29
James--Crookhelp me here...  this is at the page level rather than at the field level?  Correct?14:30
SRabbelierJames--Crook: page level, correct14:31
SRabbelierJames--Crook: but I am thinking of a gsoc project perhaps14:31
SRabbelierJames--Crook: to make it on field level14:31
SRabbelierJames--Crook: so in the model we can specify the required ACL14:31
SRabbelierJames--Crook: and pages are automagically generated14:31
James--CrookWhat you'll find is that you do need to go to field level...  and14:31
SRabbelierJames--Crook: shouldn't be too difficult, but it'd need a significant rehaul :)14:32
James--Crookthat it becomes a field attribute like the default display properties for a field.14:32
James--CrookIf you don't do that, you end up with oodles of special commands, show_for_host, show_for_people_allowed_to_see_deadlines etc...14:32
SRabbelierJames--Crook: Yup, I understand what you mean14:33
SRabbelierJames--Crook: which is why I wanted to do it this wya14:33
SRabbelierJames--Crook: just haven't had the time yet :)14:33
James--Crookand...  the permission system needs to become as simple as the one used in linux.14:33
SRabbelierJames--Crook: either a nice GSoC project, or something I can do when things settle down14:33
SRabbelierJames--Crook: our document ACL system is like that14:33
SRabbelierJames--Crook: but you know that :)14:33
James--CrookWhen you've done that, the ACL as an obstacle to wikification problem melts away to a very large extent.  At least I think it probably does.14:34
James--CrookBecause everything is totally 'regular'.14:34
James--CrookYou'll still need custom stuff to send out emails and such which is beyond 'meta-wiki'14:35
James--CrookBut I don't see a fundamental reason why at some time in the future an org page is just like any other form.14:36
SRabbelierIt already is, really14:36
SRabbelierAll editable views are just forms, nothing special about it14:37
James--CrookSRabbelier: The earlier discussion - a document IS a field, so it's a case of carrying over the same permission system to both.14:37
SRabbelierJames--Crook: I'm not sure I understand what you mean?14:38
James--CrookI was talking with Merio about Melange viewed as a 'super wiki'.  Basically an ordinary wiki is Melange with a single kind of form, and that form is fixed.  It has one field, which is a 'document' field.14:38
*** MatthewWilkes has quit IRC14:39
James--CrookSo my question I suppose is, if documents have linux like permission system, can that same code be resused for fields when we go to field level access?14:40
SRabbelierJames--Crook: it wouldn't scale14:41
SRabbelierJames--Crook: there's certain expectations built in to AppEngine14:42
SRabbelierJames--Crook: like, "don't show a whole bunch of entities in one page" :P14:42
SRabbelierJames--Crook: which is why the default list limit is 100, and only upped to 500 and 1000 for admins14:42
James--CrookSo how will quiz ever work?14:42
haoyuI have a project ideas search works :)14:43
tpb<$_> (at
SRabbelierhaoyu: that's pretty awesome :P14:44
James--Crookhaoyu: lovely.  I just used it to search for LISP and got four hits.  Excellent idea.14:44
James--CrookSRabbelier: I think quiz WOULD work, but we just have to do more in python and less in appengine.  It seems to be the secret of using appengine.  Use it to filter 'big chunks' and use python for the fine granularity.14:45
SRabbelierJames--Crook: no, that doesn't work :(14:45
SRabbelierJames--Crook: retrieving many entities _at all_ is a big no-no14:45
haoyuwell, thanks to google again :)14:45
SRabbelierJames--Crook: having to fetch >20 entities just to get a Quiz is not good14:46
James--CrookNot retrieving many entities :-)14:46
James--CrookSo, we  have to use lists?14:46
James--CrookThey count as one entity don't they?14:47
SRabbelierJames--Crook: the quiz? I thought we wanted them to be made up of multiple Questions14:47
James--CrookSure, OI agree.  And what you are saying14:48
James--Crookis that if we use appengine in the default way, we end up with one entity per question.14:48
James--Crookand that that is no good.14:48
SRabbelierwhat we could do14:49
SRabbelieris store a JSON object for example14:49
James--CrookBut we can have an entity with a list of questions in it.14:49
James--CrookJSON is another way.14:49
James--CrookIt's horrible to have to do such a workaround.14:49
SRabbelierhehe :P14:49
SRabbelierAppEngine is not made with a Wiki in mind14:49
SRabbelierthat is, a wiki like you mean14:49
James--CrookNo :-)14:50
James--CrookThe issue tracker, is that written on app engine?14:50
James--CrookOK.  I see a new danger, and a way to work around it that is counter intuitive.14:52
James--CrookWe'll be making architectural decisions that are forced on us by appengine.14:53
SRabbelierJames--Crook: no it's not, but it could be14:53
SRabbelierJames--Crook: notice how all list views are limited in how many entries they will show at a time14:53
SRabbelierJames--Crook: yes, we are doing that already :P14:53
SRabbelierJames--Crook: it's the reason why link_id's are  immutable :P14:54
James--CrookWhat we want to be working on is a system where it as cheap to ask for 2,000 field values as it is to ask for 100 records of 20 fields.14:54
SRabbelierJames--Crook: I don't think that's feasable :(14:55
James--CrookOh, but it is :-)14:55
James--CrookWith caveats.14:55
SRabbeliernot unless you do ugly things like, storing multiple entries in one Entity14:56
James--CrookWe cannot ask for 2000 random elements.14:56
SRabbelierJames--Crook: we need to be able to ask for a specific element14:56
James--CrookBut we'd like our code to look as if it had that freedom?  No?14:56
James--CrookOK.  I've convinced myself that GAE is not a barrier to a super-wiki.  Now my problem is to convince you.14:59
James--CrookSo give me a case where we want to do something, and GAE stops us.  I'll try and show why it doesn't.15:00
*** ajaksu has joined #melange15:05
SRabbelierHeh, I dunno, I just think we'd run into even more performance problems, on top of what we're already facing15:05
SRabbelierAlso, it'd be pretty much impossible to use the Datestore viewer15:05
SRabbeliersince we wouldn't have different Entity types anymore15:05
James--CrookBasically we end up more and more pushed into using bigtable for BIG objects.  Almost more like a file syetm than a database.15:06
James--CrookI think that as things stand GAE is dangerous.15:06
James--Crookthe perfomrance problems are a GAE issue.  In my view it is possible to implement soemthing like GAE where speed is more15:07
James--Crookdependent on the size of what you retrieve than on the number of elements (provided that you have an index)15:08
ajaksuSRabbelier: perfomance problems? :/15:08
SRabbelierajaksu: yes15:09
James--CrookI'm anxious that to work around this GAE issue, again and again in different place, we will more and more pollute the melange code with workarounds.  So my thinking is that (not now) but at some later stage15:09
SRabbelierJames--Crook: GAE is built to retrievie objects fast15:09
SRabbelierJames--Crook: it's not built to retrieve many objects fast15:09
James--Crookwe will need some abstraction...15:09
SRabbelierJames--Crook: which it can't, due to i'ts storing objects on different servers15:09
James--CrookHang on then...  It should be able to retrieve say 2000 objects fast?15:10
James--Crook(but we are given a 1000 object limit)15:10
SRabbelierJames--Crook: no, it's _impossible_ to do so15:10
SRabbelierJames--Crook: the more objects you retrieve, the longer the request takes15:11
SRabbelierJames--Crook: retrieving 1000 entities takes 10 seconds15:11
James--CrookNo no, I do get that.15:12
James--CrookBy 'should' I am saying 'I believe it would have been possible for Google to design a system which WAS slow for 1000 large objects, but was fast for 100 small objects.  Nothing like 10s'.  However that is not what GAE is, and we are now working around that.15:13
SRabbelierJames--Crook: I do not think it is possible15:13
James--CrookOK.  That's another disucssion.  :-)15:14
SRabbelierJames--Crook: that is, not if you want to scale to hundred thousands of entities15:14
James--CrookMmm :-)  Maybe I should offer to do that for Google ;-)15:14
SRabbelierJames--Crook: lol15:14
SRabbelierJames--Crook: you go do that ;)15:14
SRabbelierJames--Crook: on the next #appengine chat, talk with one of the devs15:14
dbentleyAre you finding app engine to be slow for 100 small objects?15:14
SRabbelierdbentley: as said, it takes 10 seconds  for 1000 256byte objects15:15
SRabbelierdbentley: I don't think it gets a lot smaller than 256 bytes ;)15:15
dbentleyTo do what operation?15:15
SRabbelierdbentley: fetch15:15
SRabbelierdbentley: just a plain fetch15:15
SRabbelierdbentley: no filtering15:15
SRabbelierdbentley: no ordering15:15
dbentleyHmm; always?15:15
SRabbelierdbentley: yes15:15
SRabbelierdbentley: always :P15:15
dbentleyIf you have a sample app that shows that behavior, I think the app engine team would be very interested; I'd happily forward.15:16
durin42SRabbelier: are they sequential keys, or random keys?15:16
durin42dbentley: if it was a bunch of random keys, isn't that a pathological case for bigtable?15:16
SRabbelierdbentley: simple, create 1000 soc/models/ User objects15:16
durin42(realizing I have no idea how the GAE datastore works over top of BT, merely guesses)15:16
SRabbelierdbentley: do a simple fetch on them to return that as JSON object15:16
James--Crookdbentley: This is so cool, that the discussion gets to this point.  I love the way collaboration works.15:17
*** durin42 has quit IRC15:20
*** Odd_Bloke has quit IRC15:20
*** Kraln has quit IRC15:20
*** scorche|sh has quit IRC15:20
*** lh has quit IRC15:20
*** krnl_ has quit IRC15:20
*** sandy|lurk has quit IRC15:20
*** ajaksu has quit IRC15:20
*** SRabbelier has quit IRC15:20
*** hpatwork has quit IRC15:20
*** lisppaste9 has quit IRC15:20
*** tansell has quit IRC15:20
*** tlsa has quit IRC15:20
*** aleb has quit IRC15:20
*** mithro has quit IRC15:20
*** r0bby_ has quit IRC15:20
*** kblin has quit IRC15:20
*** nerus has quit IRC15:20
*** Sledge has quit IRC15:21
*** scorche has quit IRC15:21
*** ajuonline has quit IRC15:21
*** danderson has quit IRC15:21
*** kangaroo has quit IRC15:21
*** summatusmentis has quit IRC15:21
*** Lennie|Food has quit IRC15:21
*** arun_ has quit IRC15:21
*** madrazr has quit IRC15:21
*** James--Crook has quit IRC15:21
*** BarryCarlyon has quit IRC15:21
*** dbentley has quit IRC15:21
*** haoyu has quit IRC15:21
*** schultmc has quit IRC15:21
*** icy has quit IRC15:21
*** arun has quit IRC15:21
*** Jeff_S has quit IRC15:21
*** uriel has quit IRC15:21
*** cwant has quit IRC15:21
*** Lennie|Food has joined #melange15:23
*** ajaksu has joined #melange15:23
*** James--Crook has joined #melange15:23
*** BarryCarlyon has joined #melange15:23
*** nerus has joined #melange15:23
*** kangaroo has joined #melange15:23
*** cwant has joined #melange15:23
*** madrazr has joined #melange15:23
*** mithro has joined #melange15:23
*** SRabbelier has joined #melange15:23
*** r0bby_ has joined #melange15:23
*** tlsa has joined #melange15:23
*** lh has joined #melange15:23
*** ajuonline has joined #melange15:23
*** arun_ has joined #melange15:23
*** krnl_ has joined #melange15:23
*** hpatwork has joined #melange15:23
*** arun has joined #melange15:23
*** haoyu has joined #melange15:23
*** tansell has joined #melange15:23
*** durin42 has joined #melange15:23
*** sets mode: +ovov Lennie|Food mithro SRabbelier durin4215:23
*** scorche|sh has joined #melange15:23
*** Odd_Bloke has joined #melange15:23
*** Kraln has joined #melange15:23
*** schultmc has joined #melange15:23
*** Sledge has joined #melange15:23
*** sandy|lurk has joined #melange15:23
*** dbentley has joined #melange15:23
*** lisppaste9 has joined #melange15:23
*** icy has joined #melange15:23
*** kblin has joined #melange15:23
*** danderson has joined #melange15:23
*** aleb has joined #melange15:23
*** scorche has joined #melange15:23
*** summatusmentis has joined #melange15:23
*** uriel has joined #melange15:23
*** Jeff_S has joined #melange15:23
*** sets mode: +o danderson15:23
SRabbelierdbentley: no, none other than the fact that a db request is allowed to take 10s most, and that you can fetch only 1000 entities at a time15:23
dbentleyWell, the fetching 1000 entities at a time is because you should be able to roll it up/sort in other ways.15:23
SRabbelierwhat do you mean?15:23
dbentleyAnd to fetch the 1001st, you need to count through 1000, which means talking to many mahcines.15:23
dbentleyLike, the way that we discussed...15:23
dbentleyDisclaimer:  This is my understand; I could be wrong.15:23
* SRabbelier nods15:23
dbentleySlow:  Give me the 1001st in this sorting.15:23
SRabbelierdbentley: it wont' even let you do that15:24
dbentleyFast:  I counted up to 1000, and the 1000th entry was "foo", so give me the next entry after "foo".15:24
SRabbelierdbentley: right, exactly15:24
*** MatthewWilkes has joined #melange15:24
dbentleyBecause then you only have to look at 1 or 2.15:24
* SRabbelier nods15:24
dbentleyCan you tell me the line that generates this query right now?15:24
SRabbelierdbentley: it's
tpb<> (at
SRabbelierdbentley: and the query I'm talking about passes an empty filter dict15:27
SRabbelierdbentley: so the resulting query is db.Query(user.User).fetch(1000)15:28
*** Lennie|Food is now known as Lennie15:31
ajaksuSRabbelier: any chances that the user model is the problem? or is the same performance present for simpler models?15:31
SRabbelierajaksu: the User model is pretyt simple :P15:32
SRabbelierajaksu: (and I tested, it takes only 256 bytes in the memcache)15:32
SRabbelieras in15:33
SRabbelierstart out with an empty memcache15:33
SRabbelierthen add 1000 user entities15:33
SRabbeliernotice that the size of the memcache increases with 256*1000 bytes15:33
ajaksuhm, so things like former_accounts in User and scope in Linkable don't influence fetch performance?15:34
SRabbelierajaksu: I highly doubt it :P15:34
SRabbelierajaksu: I'll test though15:34
ajaksuI'm thinking about happens on datastore side :)15:34
ajaksuabout *what* :)15:35
* SRabbelier nods15:36
SRabbelierajaksu: iirc it gets stored as one of them stream things15:36
SRabbelierwhat's it called again15:36
SRabbelierGoogle published an API for that15:36
ajaksuthe "Infinite item fetch generator for remote_api" cookbook recipe doesn't apply here, I guess15:39
dbentleySo, the other question is, could you do a series of items fetched -> time?  Is it linear?  Quadratic?  Etc.?15:40
dbentleyBut, yeah, my questions are clear.15:40
dbentleyI'm sorry, I have to go workon some other stuff right now (to drive more traffic to socghop! : P )  But the ability to reproduce easily is the key to making this fixable.  On some end.15:40
SRabbelierdbentley: it seems to be linear indeed15:43
SRabbelierdbentley: but I'll get you a reproducer15:43
SRabbelierajaksu: I can use a infinite fetch thingy, the problem is that requrest can't last longer than 30s15:43
SRabbelierajaksu: so we can't get more than 3k entities15:43
SRabbelierajaksu: considering we'll have 7000 users... :P15:44
ajaksuSRabbelier: "This utility function needs a function passed into it which will create a Query object and return it. If you want to be able to start from a certain point (ie, pick up where a previous run left off), it is also possible to provide a starting key"  ?15:44
SRabbelierajaksu: yeah, but we want to give a list of all users15:45
SRabbelierajaksu: for autocompletion :)15:45
SRabbelier(which is when I ran into the problem of the super-slow queryies :P)15:45
ajaksuwhat? nothing more than the user name? or the full user info?15:45
SRabbelierajaksu: user name and link id15:46
ajaksuerr... how about keeping that (tuple) list cached and update the cache with the slow infinite fetching from time to time?15:48
SRabbelierajaksu: yup, we could do that, but we'd need long-running/scheduled tasks for that15:48
SRabbelierajaksu: to so said "from time to time" updating15:49
dbentleyOh, wait, can't we do autocompletion based only on starting letter?15:49
SRabbelierdbentley: only kinda15:49
SRabbelierdbentley: I mean, we want to be able to support me typing "Rabbelier" and get srabbelier as suggested link id15:49
dbentleyOh, sure, but you don't need to fetch all the entities for that.15:49
SRabbelierdbentley: or "Hawnthorn" and get "lh" as suggested link id, based on teh fact that her name is "Leslie Hawthorn"15:49
SRabbelierdbentley: GAE doesn't have a "like" operator though :P15:50
dbentleyUgh; let's not support like.15:50
SRabbelierdbentley: i know :P15:50
SRabbelierdbentley: iterating over all entities is not fun15:50
SRabbelierdbentley: then again, it's what we want, which is why we throw the list at the user, and have AJAX do the filtering15:50
MerioSRabbelier: isn't the autocompletion doing it already (obviously if it has all the list ATM)?15:50
SRabbelierMerio: yeah, it does, but we can cache the full list15:51
SRabbelierMerio: we can't really effecitvely cache queries15:51
dbentleyBut the idea that we can actually support over all entities is close to specious...15:51
SRabbelierdbentley: what do you mean?15:51
*** James--Crook has left #melange15:52
dbentleyI just mean that what happens one year when we get 15k people, and it falls over on the client-side.15:52
dbentleyI dunno, sorry, I'm a bit distracted.  You guys chat. : )15:53
SRabbelierdbentley: I understand what you mean15:53
ajaksuit that is the performance issue, I'd offer a 'search for user' page and leave it at that :P15:53
SRabbelierajaksu: how would "search for a user" work?15:53
SRabbelierdbentley: We're probably goign to do filtering by link id15:54
ajaksua form in a separate, a simple query, user copies the name/linkid and pastes in required form...15:54
SRabbelierdbentley: so we'll only support matching "srab" to "srabbelier"15:55
ajaksudbentley: hmm, client-side with gears might be more robust? :)15:55
SRabbelierajaksu: we'd only be able to do full matches though15:55
dbentleyRight; which you can do with AppEngine...15:55
SRabbelierdbentley: yup, that's why we're probably going to stick with that :P15:55
dbentley: )15:56
LennieI also would like to know what AppError 1 is :D16:00
Lennieso that we can succesfully mail students ^_^16:00
Lenniewithout spreading it over a lot of hours :D16:00
SRabbelierme too16:00
SRabbelierdbentley: if you can poke someone about that16:00
SRabbelierdbentley: About why we get "AppError 1" when sending mails to students16:00
LennieApplication Error 116:01
Lennieto be precise :)16:01
dbentleyCan you show me a more precise error?16:01
Lennie*mail to orgs16:01
Lenniewant me to send a log?16:01
Lennieor do you have access to our logs?16:01
Lenniesocghop that is16:01
SRabbelierdbentley: that's the point, the error is really lacking16:01
dbentleyIf you can get me all the information we were presented about it, that would allow me to make the case stronger.16:02
Lenniesee pm :)16:02
LennieI pm'ed him the link to the logs :)16:03
Lenniegotta do the dishes now16:03
*** ajuonline has left #melange16:09
*** ajuonline has joined #melange16:10
ajaksuSRabbelier: but I think "from time to time" updating could use user changes as triggers... keep a 'last updated' reference somewhere and if (now - last_up) > x, refresh the cache (only check on user change, cache is fine otherwise)16:23
ajaksu(occurred to me really afk :))16:23
SRabbelierajaksu: problem with that is that users then get slow requests because we're doing stuff in the meanwhile we should really be doign in the background :P16:25
ajaksutrue, that... :)16:26
*** madrazr1 has joined #melange16:34
*** madrazr has quit IRC16:35
*** madrazr1 is now known as madrazr16:35
*** krnl_ has left #melange16:50
SRabbelierdbentley: you there by chance?17:16
*** ArthurLiu has joined #melange17:17
*** codestasher has joined #melange17:17
ArthurLiuis there a way to check the identity of an applicant mentor before accepting/rejecting him ?17:17
Lennieit is the internet17:18
Lennieyou neve know17:18
ArthurLiuall we have is the linkid, sometimes we can guess who it is and ask him through other channels but...17:18
LennieI can get you the public name17:18
Lenniebut that can be a lie too17:18
ArthurLiucould we at least have a confirmed email somewhere ?17:18
Lenniesorry can't17:19
Lennieat this moment we are not allowed to show that17:19
LennieI can check if it is a certain email address for you ^_^17:19
Lenniejust this one time17:19
SRabbelierArthurLiu: ask them to mail you17:19
SRabbelierArthurLiu: c'mon man, you can solve this without Melange ;)17:19
ArthurLiuyeah, but some went ahead and we don't know who they are...17:20
SRabbelierArthurLiu: if you want to know your mentors, have them call you on skype17:20
SRabbelierArthurLiu: and tell you what their link id is17:20
Lennieyou can find their public name though17:20<link_id>17:20
Lenniethe name will shop up on the request in the new version too17:20
Lenniebut no email17:20
ArthurLiuok thanks, that should be better17:21
Lennieyou must understand, some people dont want to give away their google account address17:21
Lenniethat's the world17:21
*** codestasher1 has joined #melange17:45
*** codestasher2 has joined #melange17:47
*** madrazr has quit IRC17:51
*** codestasher has quit IRC18:08
*** codestasher1 has quit IRC18:09
*** krnl_ has joined #melange18:14
*** dhaun has quit IRC18:20
*** mithro has quit IRC18:51
*** russellb has joined #melange19:09
*** Lennie has quit IRC19:10
russellbI'm having trouble with my organization profile.  My project (Asterisk) is listed as not having filled it out yet.  However, I filled it out minutes after finding out we were accepted.19:10
russellbNow, if I go my org profile, all the info I put in is not there, and if I fill it in and submit again, I got an error that my link ID is already in use.19:10
russellbOf course, I can't change the link ID ...19:10
russellbso ... I'm stuck.19:10
*** leifmadsen has joined #melange19:11
ajaksurussellb: any chance you used different google accounts in this process?19:13
russellbNope, I only have one.19:13
*** SRabbelier has quit IRC19:13
ajaksuhm, if nobody answers here nor on #gsoc, please file an issue at
tpbTitle: Issues - soc - Google Code (at
tpb<> (at
*** leifmadsen has left #melange19:42
*** lh has quit IRC19:44
*** hpatwork has quit IRC20:14
*** Merio has quit IRC20:28
*** MatthewWilkes has quit IRC20:49
*** allen has joined #melange21:04
*** allen has quit IRC21:26
*** allen_ has joined #melange21:26
*** codestasher2 has left #melange21:29
*** icy has left #melange22:00
*** allen_ has quit IRC22:26
*** russellb has left #melange22:37
*** allen_ has joined #melange22:40
*** lasitha_r has joined #melange22:49
*** osgeo_wolf has quit IRC23:07
*** lasitha_r has quit IRC23:12

Generated by 2.13.1 by Marius Gedminas - find it at!