| === mthaddon` is now known as mthaddon | ||
| miha | hello peoples :D | 11:01 |
|---|---|---|
| aquarius | hi miha | 11:09 |
| miha | hey, instace of NWO. no new age for you. | 11:09 |
| miha | hehe | 11:09 |
| aquarius | :) | 11:10 |
| miha | (last time "in this New World Order" was mentioned on CNN.. i guess it's official now) | 11:10 |
| === miha is now known as wolfeysi | ||
| homeasvs_ | rodrigo_, ping | 13:32 |
| rodrigo_ | homeasvs_: pong | 13:32 |
| homeasvs_ | rodrigo_, so, I have packages for n900 | 13:32 |
| homeasvs_ | rodrigo_, but | 13:32 |
| homeasvs_ | I have a cryptic erlang error on couchdb startup | 13:32 |
| homeasvs_ | http://pastebin.com/d287f37b9 | 13:32 |
| rodrigo_ | hmm, what error is it? | 13:32 |
| homeasvs_ | any idea what that's about ? | 13:32 |
| rodrigo_ | no, we'd better ask on the #couchdb channel | 13:33 |
| rodrigo_ | can you do it please? I'm out for a lunch, bbiab | 13:33 |
| homeasvs_ | sure | 13:33 |
| homeasvs_ | hm, some googling suggests I did not compile erlang with ssl | 13:33 |
| rodrigo_ | oh, yopu already did :) | 13:33 |
| homeasvs_ | that would, uhm, 'suck' to figure out now | 13:33 |
| rodrigo_ | yeah | 13:34 |
| rodrigo_ | ok, bbiab | 13:34 |
| aquarius | homeasvs_, "init terminating in do_boot" is a generic "it didn't start up properly for some reason" error :( | 13:37 |
| aquarius | homeasvs_, quite often it means "I tried to use a library and it wasn't there", which lends weight to your SSL theory | 13:37 |
| aquarius | homeasvs_, although I've had it do the same thing when spidermonkey wasn't available too | 13:37 |
| homeasvs_ | aquarius, it looks like ssl, not sure though why it was not built, investigating now | 13:38 |
| homeasvs_ | aquarius, --enable-dynamic-ssl-lib \ | 13:39 |
| homeasvs_ | --without-ssl-zlib | 13:39 |
| homeasvs_ | that's in the deb build | 13:39 |
| homeasvs_ | aquarius, any idea what exactly that then does ? | 13:39 |
| aquarius | homeasvs_, I haven't, sorry. I don't know about the build stuff; you want cardinalfang for that, I think | 13:40 |
| aquarius | #couchdb may be more helpful | 13:40 |
| aquarius | heh, in fact they are already being more helpful :) | 13:40 |
| homeasvs_ | aquarius, rodrigo_ ok, one crash further | 13:58 |
| homeasvs_ | it's so close I can smell it | 13:58 |
| homeasvs_ | aquarius, will you be getting an n900 if this works ? :) | 13:58 |
| aquarius | progress by crashing! :) | 13:58 |
| aquarius | homeasvs_, I don't like physical keyboards, otherwise I'd already be agitating for one, I think :) | 13:58 |
| homeasvs_ | aquarius, you don't have to use the physical one :) | 13:59 |
| aquarius | homeasvs_, yeah, but then I've got a phone which is half an inch thicker than it needs to be.... | 13:59 |
| homeasvs_ | aquarius, if half of that is couchdb, why does it matter ? :) | 13:59 |
| homeasvs_ | but I feel you, this phone is big in comparison to my e51 from before | 14:00 |
| * aquarius grins. I do rather like the idea of couch on the phone, I admit it | 14:00 | |
| urbanape | morning, all | 14:20 |
| rtgz | mornin.gz | 14:33 |
| rodrigo_ | homeasvs_: cool :) | 14:39 |
| homeasvs_ | sweet baby jesus | 14:42 |
| homeasvs_ | I just browsed to futon running from my phone | 14:42 |
| homeasvs_ | this calls for some food | 14:42 |
| rodrigo_ | woohoo! | 14:44 |
| === jcastro_ is now known as jcastro | ||
| dobey | hmm | 14:58 |
| CardinalFang | Desktop+ MEETING BEGINS. Say 'me' to claim a slice of the stand-up meeting, then take your turn by saying DONE/TODO/BLOCKED. | 15:01 |
| rodrigo_ | me | 15:04 |
| CardinalFang | me | 15:05 |
| aquarius | me | 15:05 |
| dobey | rodrigo_: should i comment on your blog or your facebook? I don't know any more what the cool thing to do is :) | 15:05 |
| dobey | me | 15:05 |
| rodrigo_ | dobey: up to you, I'll get notifications on both :) | 15:05 |
| rodrigo_ | dobey: the cool thing though is to do it via your Palm Pre :D | 15:06 |
| rodrigo_ | dobey: and it would be super cool if you could re-review my libubutuone branch you commented on last Friday :) | 15:07 |
| dobey | yes i will | 15:07 |
| dobey | where is everyone else? | 15:07 |
| aquarius | jblount is off ill | 15:07 |
| dobey | ChipAway, urbanape: ping | 15:07 |
| dobey | vds: ping | 15:09 |
| dobey | jblount is sick and teknico is on holiday | 15:09 |
| dobey | rodrigo_: eh, let's get this over with :) | 15:10 |
| rodrigo_ | ok | 15:10 |
| rodrigo_ | • DONE: Started adding search to contacts picker. Conboy unstable testing. XML<->HTML fixes | 15:10 |
| rodrigo_ | • TODO: Conflict resolver tool in pair tool. Look at becoming a MOTU (https://wiki.ubuntu.com/UbuntuDevelopers). Make sandy's snowy test suite work with our server (http://git.gnome.org/cgit/snowy/tree/api/tests.py). Discuss with jdo and aquarius about oauth token per app, not per machine? | 15:10 |
| rodrigo_ | • BLOCKED: no | 15:10 |
| rodrigo_ | CardinalFang: go! | 15:10 |
| urbanape | me | 15:10 |
| CardinalFang | DONE: Attachments. Finally figure it out. Played with GTG started writing desktopcouch backend storage. | 15:11 |
| CardinalFang | TODO: Still testing weird edge cases. May have to harass python-couchdb upstream. | 15:11 |
| CardinalFang | BLOCKED: None | 15:11 |
| CardinalFang | aquarius. | 15:11 |
| aquarius | ⚀ DONE: talk to thisfred and vds about sequence numbers etc; talk to #couchdb about sequence numbers; review duty; decide on lxml, not BeautifulSoup, for html parsing with rodrigo | 15:11 |
| aquarius | ⚁ TODO: make tomboy first-sync experience nicer; continue work on desktopcouch developer docs; write up things learned at UDS/sprint; work with rodrigo on Music Store, much more music store architecture planning; step-by-step guide to what happens during contact sync | 15:11 |
| aquarius | ⚂ BLOCKED: | 15:11 |
| aquarius | dobey: go | 15:11 |
| dobey | ☺ DONE: Reviews, SSO call, Triage | 15:11 |
| dobey | ☹ TODO: Homework, Review/Fix #479375, Finish Backporting, Triage, Prepare releases/SRUs | 15:11 |
| dobey | ☹ BLCK: None. | 15:11 |
| dobey | urbanape: rock on | 15:11 |
| urbanape | DONE: Got ubuntuone-servers branch back from the dead, handed off to jblount for some CSS lovin | 15:11 |
| urbanape | TODO: Get it landed. Move on to other bugs and Bindwood. | 15:11 |
| urbanape | EOM? | 15:11 |
| urbanape | BLOCK: None | 15:11 |
| rodrigo_ | sandy|lurk, aquarius: around here? | 15:17 |
| aquarius | ya | 15:17 |
| sandy|lurk | yup | 15:17 |
| rodrigo_ | ok, so about storing HTML vs XML on our database, sandy|lurk, could you please share your concerns with aquarius? | 15:18 |
| sandy|lurk | sure | 15:18 |
| sandy|lurk | so currently there's this bug in the XML<->HTML parser | 15:18 |
| sandy|lurk | which messes up bulleted lists a bit | 15:18 |
| aquarius | yes | 15:18 |
| sandy|lurk | at first I thought, okay, no big deal, should only affect users who edit notes in U1 | 15:18 |
| sandy|lurk | but then I realized that actually, you store HTML, and the conversion happens on each sync | 15:19 |
| sandy|lurk | well, this means anyone syncing experiences the bug | 15:19 |
| sandy|lurk | so even though I'm sure this will be fixed soon | 15:19 |
| sandy|lurk | this sort of thing could easily come up again | 15:19 |
| aquarius | yes. | 15:19 |
| aquarius | Your proposed alternative is, I imagine, store tomboy's XML in the DB, yes? | 15:19 |
| sandy|lurk | so I'd advocate storing the native XML format of the REST API instead | 15:19 |
| sandy|lurk | yup | 15:19 |
| aquarius | my problem with doing that is that we have to have a working html-to-xml converter for the web UI anyway. | 15:20 |
| rodrigo_ | aquarius: well, it makes sense, since it's the format used by tomboy, conboy, tomdroid, etc | 15:20 |
| sandy|lurk | as I explained to rodrigo, I woudl guess that most users using Tomboy+U1 are using it for syncing, then a subset uses it for viewing too, and then a smaller subset uses it for editing | 15:20 |
| rodrigo_ | aquarius: or an editor that knows about the XML format, would that be too hard? | 15:20 |
| homeasvs_ | rodrigo_, check my facebook :) | 15:20 |
| sandy|lurk | aquarius: yeah, I sympathize with that issue | 15:20 |
| aquarius | rodrigo_, yep, it's too hard. I tried that. Without writing a widget completely, completely from scratch, whcih is horrific pain | 15:20 |
| sandy|lurk | yes, the rich editors all use HTML | 15:21 |
| sandy|lurk | that makes sense | 15:21 |
| sandy|lurk | aquarius: you could have a cache, of course | 15:21 |
| aquarius | sandy|lurk, I'm trying to avoid the web UI falling behind if there's a problem. This way, if there's a problem we see it immediately and can fix it immediately | 15:21 |
| sandy|lurk | generate the HTML only when the user actually views the note...slowish the first time | 15:21 |
| aquarius | cache? | 15:21 |
| sandy|lurk | yeah, store both | 15:21 |
| aquarius | yeah, but we have to back-translate | 15:21 |
| aquarius | from arbitrary random HTML to tomboy XML | 15:21 |
| sandy|lurk | aquarius: only if they edit | 15:22 |
| sandy|lurk | aquarius: I understand that this way you see bugs faster | 15:22 |
| sandy|lurk | but users see bugs immediately that they woudl not expect | 15:22 |
| aquarius | yeah. I'm not completely averse to the idea of storing tomboy XML in the DB, I admit | 15:22 |
| sandy|lurk | I would not expect simply syncing my notes to modify them in any way | 15:22 |
| sandy|lurk | (as a user) | 15:22 |
| rodrigo_ | storing both formats might be ok, I guess | 15:23 |
| aquarius | oh, I agree | 15:23 |
| sandy|lurk | yeah, I don't know what your space constraints are | 15:23 |
| rodrigo_ | homeasvs_: in a minute... | 15:23 |
| sandy|lurk | text is small but doubling the storage could hurt, I dunno | 15:23 |
| aquarius | space isn't a problem; my issue is that storing the same data twice is a flat-out invitation for them to get out of sync :) | 15:24 |
| sandy|lurk | but ideally, storing both would be my favorite soution | 15:24 |
| aquarius | if they're translatable, then you don't need to store them twice. If they're not translatable, then the web UI is irretrievably broken anyway | 15:24 |
| rodrigo_ | yeah, and whenever you sync a note from tomboy, we need to convert, and the same when editing on the web | 15:24 |
| rodrigo_ | which gives us the same problem, if there are new tags in the tomboy xml not understood by the converter | 15:24 |
| sandy|lurk | well you already convert every note during a sync | 15:25 |
| aquarius | my original intention was, as above, to make it so if it broke then it'd break for everyone and we'd notice and fix it fast | 15:25 |
| rodrigo_ | yeah | 15:25 |
| sandy|lurk | even though most users will nver usethe HTML | 15:25 |
| rodrigo_ | but if the converter is buggy, we have the same problem | 15:25 |
| sandy|lurk | aquarius: yeah, I understand that | 15:25 |
| sandy|lurk | but that seems pretty risky to me | 15:25 |
| aquarius | but I am starting to think that the better approach is to have it so when it breaks, it doesn't necessarily break everything | 15:25 |
| aquarius | so at least a subset of users won't experience the problem. | 15:25 |
| sandy|lurk | I think users using different editors on the same file may expect some weirdness | 15:26 |
| sandy|lurk | if I use Word and OO.o I expect formatting issues | 15:26 |
| sandy|lurk | ditto Tomboy and U1's note editor | 15:26 |
| rodrigo_ | sandy|lurk: you mentioned some time ago that plugins would use their own XML markup if needed, right? | 15:26 |
| sandy|lurk | rodrigo_: yeah, many plugins add new XML tags | 15:26 |
| sandy|lurk | if your converter doesn't know about them today, I don't know what you guys do | 15:27 |
| rodrigo_ | homeasvs_: yay! | 15:27 |
| sandy|lurk | if I write a custom add-in | 15:27 |
| sandy|lurk | that only I know about | 15:27 |
| sandy|lurk | and I don't share it | 15:27 |
| sandy|lurk | and it adds a new XML tag | 15:27 |
| sandy|lurk | and then I sync to U1 and it gets lost even though I don't use the editor | 15:27 |
| sandy|lurk | I'd be sad | 15:27 |
| aquarius | it should, in theory, handle it (<sandy> gets converted to <span class="sandy"> and then gets converted back) | 15:27 |
| sandy|lurk | aquarius: ah, cool :-_) | 15:27 |
| homeasvs_ | rodrigo_, I'll see about getting my packages done properly, cleaning up, and putting them somewhere | 15:27 |
| aquarius | I did try to not have it throw away information ;) | 15:27 |
| homeasvs_ | rodrigo_, then next step would be desktopcouch stuff | 15:27 |
| sandy|lurk | aquarius: thanks | 15:28 |
| sandy|lurk | so then the only real concern is converter bugs | 15:28 |
| aquarius | the current parser doesn't eat indented notes because it throws away information, it eats indented notes because it's appending children to the wrong place in the tree | 15:28 |
| sandy|lurk | if the bug was limited to the editor it would just be annoying | 15:29 |
| sandy|lurk | since it's not, it's critical | 15:29 |
| rodrigo_ | sandy|lurk: yay, I marked it as such | 15:29 |
| sandy|lurk | so a bug in the converter becomes a much bigger deal all of the sudden | 15:29 |
| aquarius | yeah, so it comes down to a straight-up choice between 1. converter bugs get noticed quickly because they break everyone, and 2. converter bugs only affect a subset of users but are less likely to get fixed when they do happen | 15:29 |
| sandy|lurk | well maybe I can help with 2) by providing good complex test notes | 15:30 |
| aquarius | I was leaning towards 1 (that's why we store HTML), but I am now leaning towards 2, with the caveat that we have to convert everyone's existing notes :) | 15:30 |
| sandy|lurk | I assume you guys have unit tests | 15:30 |
| rodrigo_ | homeasvs_: desktopcouch should be easier, I think, since python modules should be easy, right? | 15:30 |
| sandy|lurk | so better note data to test could help with 2 | 15:30 |
| aquarius | good complex test notes would be good -- we do have unit tests for the converter, but they're not totally comprehensive (that's how we missed the nested-lists bug :)) | 15:30 |
| sandy|lurk | cool, I'll try to get to that before next week then | 15:30 |
| rodrigo_ | sandy|lurk: we have unit tests, but they don't test everything, so we're adding them | 15:31 |
| aquarius | do you have a set of complex notes for tomboy's own unit tests? | 15:31 |
| sandy|lurk | so Tomboy sucks wrt unit testing | 15:31 |
| sandy|lurk | but I do have complex notes in my own collection for ad-hoc testing | 15:31 |
| * aquarius grins | 15:31 | |
| aquarius | us having them would help quite a lot :) | 15:32 |
| homeasvs_ | rodrigo_, yeah | 15:32 |
| sandy|lurk | when we were rewriting printing we came up with some fun gnarly notes to test with | 15:32 |
| homeasvs_ | rodrigo_, I guess it will depend on whether we need ubuntuone-storage-protocol | 15:32 |
| aquarius | we need to have a working converter first, though, so we can convert the stored format for current notes | 15:32 |
| sandy|lurk | okay, I've got to run...employee meeting starting | 15:32 |
| aquarius | sandy|lurk, cheers! | 15:32 |
| rodrigo_ | homeasvs_: that's for file sharing, so I guess not | 15:32 |
| homeasvs_ | but even then, should still be doable. Just want to clean up the current mess first before moving on, since I will probably need to pull in some couchdb patches as well for oauth | 15:32 |
| sandy|lurk | I also want to talk to you guys about note attachments at some point | 15:32 |
| homeasvs_ | rodrigo_, right, but I want to easily pair with u1 on the phone | 15:33 |
| aquarius | homeasvs_, you don't need any ubuntu one stuff at all for desktopcouch. | 15:33 |
| rodrigo_ | homeasvs_: ah, ok | 15:33 |
| rodrigo_ | go homeasvs_ go! :D | 15:33 |
| aquarius | homeasvs_, you need a little bit of it in order to replicate couch to u1, but not much (just enough to get the oauth tokens) | 15:33 |
| rodrigo_ | sandy|lurk: are attachments supported already in tomboy? | 15:33 |
| sandy|lurk | ie, how should we modify the REST API to support notes having files attached to them (images for example) | 15:33 |
| rodrigo_ | sandy|lurk: ah ok | 15:33 |
| aquarius | sandy|lurk, that sounds like it'll be a fun conversation :) | 15:33 |
| sandy|lurk | rodrigo_: no, but I was working on it the other day | 15:34 |
| aquarius | sandy|lurk, ah! ok, good, I thought it was going to be a kicking for us about not supporting them ;) | 15:34 |
| rodrigo_ | sandy|lurk: for images, I guess we need them | 15:34 |
| aquarius | sandy|lurk, happy to think about that, definitely :) | 15:34 |
| sandy|lurk | cool, thanks for the chat guys | 15:34 |
| rodrigo_ | from our POV it should be easy, since we can store attachments with the couchdb documents | 15:34 |
| aquarius | yeah, as long as no-one attaches an iso image to a note or something ;) | 15:34 |
| rodrigo_ | sandy|lurk: thanks to you :D | 15:34 |
| rodrigo_ | aquarius: :) | 15:35 |
| rodrigo_ | homeasvs_: oh, you mentioned the sdk repo in some facebook message, where is that? | 15:35 |
| rodrigo_ | homeasvs_: I haven't seen anything about it when looking for repos | 15:35 |
| homeasvs_ | rodrigo_, it's the one that is in your scratchbox | 15:36 |
| homeasvs_ | deb http://repository.maemo.org/ fremantle/sdk free non-free | 15:36 |
| rodrigo_ | homeasvs_: ah | 15:36 |
| aquarius | rodrigo_, so I think that we should 1. get the converter working 2. change the back-end storage format to tomboy XML 3. convert all the currently-stored notes | 15:37 |
| rodrigo_ | aquarius: step by step, I'll work on 1 first :) | 15:37 |
| mandel | CardinalFang: ping | 15:39 |
| CardinalFang | mandel, hi | 15:39 |
| mandel | CardinalFang: hello, did u see my message regarding the record_id branch? | 15:40 |
| mandel | CardinalFang, all that conversation about the if statement and at the end I used to wrong logic <embarrassed> | 15:41 |
| CardinalFang | mandel, I haven't seen that message yet, no. | 15:41 |
| mandel | https://code.edge.launchpad.net/~mandel/desktopcouch/record_id/+merge/15424 | 15:41 |
| mandel | CardinalFang, last message | 15:41 |
| mandel | CardinalFang, guillermo just changed to approved, but the code will not work | 15:42 |
| verterok | mandel, CardinalFang sorry :/ | 15:42 |
| verterok | it has 2 approves... | 15:42 |
| verterok | mandel: CardinalFang: should I cange it back to Needs review/reject? | 15:43 |
| sandy|lurk | rodrigo_: here's a test note that was great for printing, though it only goes two-levels deep for bulleted lists: http://bugzilla-attachments.gnome.org/attachment.cgi?id=127654 | 15:43 |
| mandel | verterok, CardinalFang, I would, you can blame me, it was my fault :P | 15:43 |
| verterok | mandel: should I change it to needs review? | 15:43 |
| mandel | verterok, I would | 15:44 |
| verterok | ok, done | 15:44 |
| verterok | mandel: thanks for pinting this out :) | 15:44 |
| mandel | verterok, no problem, the new version in the branch has the correct logic plus a test to avoid any other problems like this | 15:44 |
| mandel | verterok, totalmente mi culpa ;) | 15:45 |
| verterok | mandel: ooh, I see. you should "resubmit" the branch ;-) | 15:45 |
| mandel | verterok, will do | 15:45 |
| mandel | verterok, I just have to do a second merge proposal? | 15:46 |
| verterok | mandel: no, from the merge proposal pag you can resubmit it, and launchpad creates a new merge proposal that superseeds the current | 15:47 |
| CardinalFang | verterok, mandel, usually it is best to push a new branch, but there's no reason to bother now. Everyone who cares is here listening and knows what's going on. | 15:47 |
| CardinalFang | mandel, there's no red tape here. Let's worry about the bug instead. | 15:47 |
| mandel | CardinalFang, so I just do a superseed merge, ok | 15:48 |
| CardinalFang | verterok, (I don't think one can resubmit a merged branch. I may be wrong.) | 15:48 |
| verterok | CardinalFang: it's merged? | 15:49 |
| CardinalFang | Yes. | 15:49 |
| verterok | ooh, that's a different story | 15:49 |
| verterok | mandel: ^ | 15:49 |
| CardinalFang | mandel, just push it as you intended to. It's all good. | 15:49 |
| mandel | CardinalFang, verterok, so I do not do anything, right? | 15:50 |
| mandel | CardinalFang, verterok, I feel kind of stupid for that error... | 15:50 |
| verterok | mandel: what CardinalFang said | 15:51 |
| mandel | ok | 15:51 |
| verterok | mandel: no worries, I reviewed the branch and didn't see it :/ | 15:51 |
| mandel | CardinalFang, I have pushed a new version of the branch with the changes + test | 15:55 |
| mandel | CardinalFang, although I can always push to a diff branch to make it explicit | 15:55 |
| CardinalFang | mandel, there's a problem with your test. I'm adding a comment. | 16:06 |
| CardinalFang | mandel, comment added. Needs fixing. | 16:15 |
| CardinalFang | https://code.edge.launchpad.net/~mandel/desktopcouch/record_id/+merge/15753 | 16:15 |
| mandel | CardinalFang, cool I sort it out right away, I di dnot know that :D | 16:17 |
| mandel | CardinalFang, changed made, sorry for the pain in the ass | 16:31 |
| urbanape | aquarius: you aboot? | 17:47 |
| aquarius | otp | 17:47 |
| urbanape | k | 17:47 |
| urbanape | also for CardinalFang and thisfred: was thinking more about the way we'll handle delete in the future. If clients push content that has been seen before, presumably we'll need to provide a recognizable response so the client knows it can deal with the record properly. | 17:48 |
| urbanape | "Hi, here's this bookmark." | 17:49 |
| urbanape | "Bah, our user deleted that weeks ago. Begone!" | 17:49 |
| urbanape | what HTTP response code would you use? Bad Request? | 17:49 |
| urbanape | 406 Not Acceptable? | 17:50 |
| urbanape | 410 Gone | 17:50 |
| urbanape | I think 410 matches that story. | 17:51 |
| rmcbride | 411 "highly embarrassing"? | 17:51 |
| * rmcbride makes up HTTP response codes at randome | 17:51 | |
| CardinalFang | urbanape, I think upstream couchdb will be a better place to handle that, and I bet they will have something already planned. Maybe jan____ knows. | 17:51 |
| urbanape | do you know off-hand what the conditions will be for determining identity? presumably the couch _id, right? | 17:52 |
| urbanape | software is hard. let's go shopping. | 17:53 |
| CardinalFang | jan____, without consulting #couchdb at all, we've been tossing around ideas for understanding and dealing with documents that are expunged from a database, so that replication doesn't reintroduce them. | 17:56 |
| urbanape | jan____: particularly when Couch isn't the primary storage, but is the replication medium. | 17:57 |
| thisfred | again: replication does not reintroduce deleted documents... | 17:58 |
| CardinalFang | Ah. | 17:58 |
| urbanape | thisfred: it's not about the replication | 17:58 |
| thisfred | right | 17:59 |
| CardinalFang | So it is a positive event. Sorry. | 17:59 |
| urbanape | in Bindwood | 17:59 |
| urbanape | so that other clients don't reintroduce them | 17:59 |
| urbanape | (not replication, correct) | 17:59 |
| thisfred | I think the way to solve this is to monitor _changes, and act accordingly | 18:00 |
| urbanape | yes, but the question is what to monitor in changes? There has to be a there there to monitor | 18:00 |
| urbanape | if deletions show up as a document in the _changes feed, we're fine | 18:00 |
| thisfred | urbanape: ah, do delete events not show up? | 18:00 |
| thisfred | that would be an unfortunate omission | 18:00 |
| urbanape | I don't know, we don't delete documents | 18:00 |
| urbanape | lessee | 18:01 |
| thisfred | right, so we don't have the problem yet, or do we? | 18:01 |
| urbanape | yes, they show up in changes | 18:02 |
| urbanape | with a deleted flag | 18:02 |
| urbanape | yay, we're done. | 18:02 |
| urbanape | champagne and caviar, for everyone! | 18:02 |
| thisfred | Can I have double champagne if I pass on the caviar? :) | 18:02 |
| urbanape | no. Eat your fish eggs and like 'em. | 18:03 |
| * thisfred mopes | 18:03 | |
| urbanape | CardinalFang: we (at least Bindwood) have no problems in this regard any longer. | 18:03 |
| urbanape | forget I ever ranted and/or raved. | 18:03 |
| urbanape | we'll actually just do delete documents and pull those changes from _changes when we normally poll. | 18:04 |
| aquarius | urbanape, yo | 18:05 |
| urbanape | aquarius: never mind, we solved it. | 18:05 |
| aquarius | urbanape, oh, good. I shall read the scrollback :P | 18:06 |
| urbanape | summary: deleted docs show up in _changes with a deleted: True field in the record. | 18:08 |
| urbanape | so we don't need to tag records with deleted, now that we're polling _changes. We can just tell Couch to delete the doc, and it'll show up for other clients in _changes. | 18:09 |
| urbanape | We don't need to wait for history. | 18:09 |
| aquarius | as long as bindwood polls _changes, yeah. Excellent. That's just excellent, is what that is | 18:13 |
| aquarius | now all we need is chromium support for shelling out and then I can have my bookmarks back again :P | 18:14 |
| urbanape | yes, I need to hang out in #chromium some more. | 18:25 |
| urbanape | I just think it might be beyond their scope. | 18:26 |
| urbanape | Expeciallly in light of Chrome OS. | 18:26 |
| urbanape | "What would you shell out to? There is only the browser. What would you gain from an environment (in an env sense) that is only the browser?" | 18:27 |
| urbanape | and I don't think they want to entertain the notion of extensions that only run on "desktop OSes" | 18:28 |
| dobey | NOT A DESKTOP | 18:30 |
| dobey | kthx | 18:31 |
| dobey | :) | 18:32 |
| urbanape | your pedantry is quaint, but I don't think you're going to convince the world. | 18:32 |
| urbanape | I'm gonna start just calling everything jeejahs. | 18:34 |
| dobey | desktop jihad | 18:34 |
| dobey | convert or die :) | 18:34 |
| dobey | can we not get bookmarks and passwords out of chrome? | 18:35 |
| dobey | if not, it sounds like an antitrust suit waiting to happen :) | 18:36 |
| urbanape | we can get bookmarks. Not sure about passwords. | 18:37 |
| urbanape | the problem is we can't shell out to get our dbus couch port or get at our oauth tokens. | 18:37 |
| dobey | oh we can't run python? | 18:38 |
| urbanape | we can't shell out period. | 18:38 |
| urbanape | we can't run an external process | 18:38 |
| dobey | yes, that's what i was saying | 18:39 |
| dobey | do they provide some way to access local disk data from internals? | 18:39 |
| dobey | (please don't say "HTML5 storage") | 18:39 |
| urbanape | I believe you can get to file:/// resources, but I don't think there's a way to get to $USER from javascript, so per-user files would be hard to do. | 18:40 |
| urbanape | still learning some of this stuff, so it might be possible. | 18:40 |
| dobey | and you probably can't do things like read/write from/to unix sockets | 18:41 |
| urbanape | hmm, you can use NPAPI for "legacy" stuff. | 18:41 |
| dobey | what the heck is "legacy"? | 18:41 |
| dobey | certainly not a 2TB hard disk | 18:41 |
| urbanape | http://code.google.com/chrome/extensions/npapi.html | 18:42 |
| aquarius | "legacy": "anything that we don't explicitly provide as an extension API" | 18:42 |
| dobey | "anything that isn't google docs" | 18:45 |
| CardinalFang | aquarius, thisfred, mandel, https://code.edge.launchpad.net/~cmiller/desktopcouch/attachments/+merge/15765 | 19:00 |
| CardinalFang | Review? Complain? | 19:00 |
| * CardinalFang acquires comestibles. | 19:00 | |
| aquarius | CardinalFang, why block overwriting of an existing attachment? to stop people doing it by accident? | 19:04 |
| CardinalFang | aquarius, Yes. I want anything that destructive to be explicit. | 19:14 |
| aquarius | but record["foo"] = "bar" is destructive too, no? even if record["foo"] == "quux" | 19:14 |
| aquarius | and changes aren't *saved* | 19:14 |
| rtgz_ | Just a thing I wanted to ask some days ago. Where does the diff containing debian/ directory come from for ubuntuone-client lp trunk? | 19:15 |
| CardinalFang | aquarius, They may understand index or key assignments, but the verb "attach" doesn't convey "be sure not to have the same name already, or you lose". | 19:17 |
| CardinalFang | replace() ? | 19:17 |
| aquarius | I see your point. | 19:17 |
| aquarius | don;t want replace() since most of the time you're not replacing anything | 19:18 |
| aquarius | yeah, maybe you're right. :) | 19:18 |
| mandel | CardinalFang, looking at it right now | 19:39 |
| mandel | CardinalFang: what about using an update rather than detach, attach??? | 19:44 |
| thisfred | CardinalFang: looks good to me. Just one questions: Why do we not want CouchDB to provide ids for documents? | 19:46 |
| CardinalFang | thisfred, they say it's a bad idea. | 19:46 |
| thisfred | really? | 19:47 |
| mandel | thisfred, yes, it should be done in the client side, I had to do the same in my code | 19:48 |
| thisfred | Is the algorithm broken, or are there collission issues? | 19:48 |
| CardinalFang | From python-couchdb cleint.py : """Note that it is generally better to avoid the `create()` method and instead generate document IDs on the client side. This is due to the fact that the underlying HTTP ``POST`` method is not idempotent, and an automatic retry due to a problem somewhere on the networking stack may cause multiple documents being created in the database.""" | 19:48 |
| thisfred | anyway, that's just curiosity. If it's a better idea not to, I'll change the server side code not to do it either | 19:49 |
| thisfred | CardinalFang: ah yes, that makes sense | 19:49 |
| thisfred | Not having transactions is the price we pay | 19:50 |
| CardinalFang | thisfred, I changed the server class, "put record". | 19:50 |
| mandel | thisfred, adding to what CardinalFang said: http://www.atypical.net/archive/2009/05/12/couchdb-090-bulk-document-post-performance | 19:51 |
| thisfred | CardinalFang: right, should be enough, just checking that I'm not putting anything else bypassing the API | 19:51 |
| thisfred | like users | 19:51 |
| urbanape | hmm, I was just about to noodle on how bindwood could drop the FF-generated UUIDs in favor of getting the next id from Couch and use that as the unifying foreign key. | 19:51 |
| urbanape | or am I misunderstanding that? Are they suggesting that clients generate ids for documents? | 19:52 |
| mandel | urbanape, yes, from my understanding, the client should generate the id | 19:53 |
| urbanape | even on id-less PUTs? | 19:54 |
| mandel | CardinalFang, got a question, why do you have a try block in list_attachments?? | 19:55 |
| CardinalFang | urbanape, Yes. HTTP isn't smart enough. If you're going to be stateless, you have to be able to handle when you're redirected several times. A (e.g.) proxy between you could PUT more than once, and that would cause problems. | 19:55 |
| urbanape | CardinalFang: fair enough, I suppose | 19:56 |
| CardinalFang | mandel, good find. No reason, now. I was creating _attachments somewhere else. Now I make it at __init__(), and so I can be sure it exists. I'll remove it. | 19:57 |
| mandel | CardinalFang: in the detach, you will have a keyword error when poping, do you want to give the default message or put your own? Maybe "not existent attachment or empty" or something of the kind | 20:00 |
| mandel | CardinalFang, of course when working with an empty dict or none existing keyword, but I'm sure you knew what I meant ;) | 20:01 |
| CardinalFang | mandel, Okay, I added more descriptive KeyErrors. Thank you. | 20:11 |
| mandel | CardinalFang, One last thing, if you are going to generate the id in the client side, why not moving that to the record constructor when the record_id or data["_id"] is none. | 20:17 |
| mandel | CardinalFang, I know it is less lazy, but from my point of view it looks nicer | 20:18 |
| mandel | CardinalFang, one stored the other one takes care of the id, but this is just my opinion | 20:18 |
| CardinalFang | mandel, I want to keep as much as possible like the old behavior. | 20:19 |
| mandel | CardinalFang, ok, I understand. Then everything looks great! | 20:20 |
| mandel | CardinalFang, I'l use it as soon as it is merged, thanks! | 20:21 |
| BlackPho | Hello | 21:03 |
| CardinalFang | Phở? Yum. | 21:04 |
| === verterok is now known as verterok|brb | ||
| === verterok|brb is now known as verterok | ||
| kjoller | I have just gotten a (well, two) share-e-mails, but when I press the link, I get a "Something has gone wrong (500)". Nothing is available through the Ubuntu One folder either. | 21:35 |
| verterok | kjoller: hi, yes it's a known issue, it will be fixed real soon | 21:39 |
| kjoller | verterok: ok, great, thanks | 21:39 |
| statik | verterok, pfibiger: do we need a rollout before sharing is fixed? | 21:54 |
| statik | i am so woefully behind on things | 21:55 |
| dobey | statik: "A plan is just a list of things that don't get done." | 21:56 |
| verterok | statik: yes, the fix already landed | 21:56 |
| statik | cool, thanks | 21:57 |
| chatZilla | how do i relocate my files on ubuntuone to a different folder or rename a folder? | 22:16 |
| homeasvs_ | sweet, python-couchdb is now also working on maemo | 22:31 |
| homeasvs_ | and my commandline app runs too | 22:31 |
| homeasvs_ | next up, a gui | 22:31 |
| jan____ | CardinalFang: urbanape: ping me tomorrow CST time for discussion, sorry! | 22:33 |
| === sandy_ is now known as sandy|lurk | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!