00:02:52  * tilgovijoined
00:45:00  * d2dchatquit (Remote host closed the connection)
01:44:41  * IrishGringoquit (Quit: ChatZilla 0.9.90 [Firefox 21.0/20130511120803])
01:54:02  * tilgoviquit (Ping timeout: 246 seconds)
02:07:20  * cliffanojoined
02:19:33  <cliffano>JasonSmith_: I have a couch replication rule with source public npm registry on iriscouch... i'm seeing GET request that gets larger and larger, e.g. https://gist.github.com/cliffano/5726639 ... do you know if this is normal?
02:20:41  <cliffano>JasonSmith_: I'm sitting behind a proxy that limits the GET request size, and truncates the outbound request resulting in CouchDB to respond with error 400 bad request
02:22:21  * Slyjoined
02:30:15  <cliffano>JasonSmith_: how often does iriscouch compact the npm registry database?
02:53:51  * JasonSmith_changed nick to JasonSmith
02:54:09  <JasonSmith>cliffano: Interesting, I will take a look
03:02:11  <cliffano>JasonSmith: i have compacted the registry on my end... so i wonder if it's the main registry that might still keep those revisions
03:03:02  <JasonSmith>cliffano: Right, every CouchDB document has a change timeline, very similar to Git or Subversion
03:03:17  <JasonSmith>By default, CouchDB is set to store 1,000 revisions of history
03:03:50  <JasonSmith>Unlike Git, CouchDB does not store the old data, but only fact that today's data evolved from yesterday's data
03:04:28  <JasonSmith>When it replicates, it fetches changes for only the missing part of the time line
03:04:43  <JasonSmith>in other words, a big GET url can happen naturally
03:09:38  * Slyquit (Remote host closed the connection)
03:11:27  <cliffano>JasonSmith: do you suspect that those revisions include/exclude the compacted-away ones?
03:11:46  <JasonSmith>cliffano: Actually compaction does not affect this stuff
03:12:42  <cliffano>JasonSmith: ah i see.. was speculating about compaction from reading touchdb-ios issue which limits the number of atts_since param value https://github.com/couchbaselabs/TouchDB-iOS/issues/90
03:13:01  <JasonSmith>Yes, I made the exact same assumption in the past
03:13:35  <JasonSmith>if you imagine a CouchDB document evolving, conceptually your file has [rev1] and then later [rev2 <rev1>] and then maybe [rev3, <rev2, rev1>]
03:13:41  <cliffano>and this https://groups.google.com/forum/?fromgroups#!topic/mobile-couchbase/e4L7sRIPtwo made me think that couchdb revision tree could be trimmed down... so i guess the revision tree might have 1000 legitimate revisions
03:13:47  <JasonSmith>Stuff in <> is the ids of the old versions
03:14:16  <JasonSmith>compaction will delete your rev1 and rev2 *data* but the rev3 in the example still has the ids in its own metadata
03:14:27  <JasonSmith>Right, you can configure CouchDB to limit that
03:14:59  <JasonSmith>if you set a limit to 1 for example, then CouchdB would never remember document histories, so any time you replicated, it would be forced to assume every document is a conflict (no common ancestry)
03:15:06  <JasonSmith>It's really very much like Git
04:10:15  <cliffano>JasonSmith: thanks for checking and the confirmation :) (sorry, was distracted afk for a while)
04:10:31  <JasonSmith>cliffano: You're welcome
04:14:50  * tilgovijoined
04:46:02  * tilgoviquit (Remote host closed the connection)
06:14:48  <cliffano>JasonSmith: tried changing the limit to 50 on my end, but i still see the same error, so it must be based on the limit setting on the source (public npm registry)... do you know why couch doesn't use POST instead of GET in this case?
06:15:29  <JasonSmith>cliffano: If I recall, it will switch to POST if the length grows beyond a maximum
06:15:36  <JasonSmith>Are you doing this at work, or for personal use?
06:15:49  <cliffano>work
06:16:16  <JasonSmith>For what it's worth, we actually sell dedicated, private npm replicas with all this taken care of, in any cloud data center you need
06:16:25  <JasonSmith>for a small fraction of the cost of an engineer's salary
06:16:43  <JasonSmith>OTOH I am totally happy for people to keep DIY replicas too
06:16:52  <JasonSmith>Just saying, if you have lots of balls in the air... :)
06:17:36  * JasonSmithgoogles cliffano
06:17:44  <JasonSmith>Hey are you in Melbourne? I was just there!
06:18:16  <JasonSmith>https://twitter.com/_jhs/status/311604852138594304
06:18:23  <cliffano>JasonSmith: yep, we've discussed about iriscouch & cloudant :) unfortunately it's not a simple call to make for us (not just the dev team :p)
06:19:03  <JasonSmith>cliffano: Yes, totally. Well, I'll speak my peace and then not bother you about it.
06:19:34  <JasonSmith> is actually an "npm" product, not simply a "couchdb" product. We guarantee not couchdb, but a full up, up-to-date npm registry, you can even push private packages to it.
06:19:38  <JasonSmith>end of pitch :)
06:20:48  <cliffano>JasonSmith: i will definitely mention it again if we're completely stuck...
06:22:00  <JasonSmith>I am looking for the bugfix related to this, I remember there was a bug with large GET sizes
06:23:09  <JasonSmith>cliffano: So are you saying that your GET requests cannot penetrate your proxy?
06:29:23  * Samuel_Roldanjoined
06:38:07  * cliffanoquit (Ping timeout: 250 seconds)
06:38:59  * cliffanojoined
07:05:44  * Samuel_Roldanquit (Quit: Samuel_Roldan)
07:25:32  * tilgovijoined
07:40:05  * Samuel_Roldanjoined
07:50:10  * marfarmaquit (Read error: Connection reset by peer)
07:50:57  * marfarmajoined
07:54:21  * cliffanoquit (Quit: Page closed)
08:43:02  * Samuel_Roldanquit (Quit: Samuel_Roldan)
08:49:32  * ebrightjoined
09:17:21  * julianduquequit (Ping timeout: 248 seconds)
09:41:20  * julianduquejoined
11:12:24  * julianduquequit (Ping timeout: 240 seconds)
11:20:21  * ramitosjoined
11:39:16  * tilgoviquit (Ping timeout: 260 seconds)
12:00:55  * julianduquejoined
12:39:47  * Slyjoined
13:01:34  * marfarma_joined
13:03:01  * marfarmaquit (Ping timeout: 240 seconds)
13:50:21  * Samuel_Roldanjoined
14:10:40  * Samuel_Roldanquit (Quit: Samuel_Roldan)
14:58:48  * ebrightquit (Ping timeout: 256 seconds)
17:31:49  * Samuel_Roldanjoined
18:51:04  * Slyquit (Remote host closed the connection)
18:53:47  * Slyjoined
18:56:09  * marfarma_quit (Read error: Connection reset by peer)
18:56:46  * marfarma_joined
19:16:10  * tilgovijoined
19:56:53  * Slyquit (Ping timeout: 248 seconds)
20:11:52  * marfarmajoined
20:17:08  * Samuel_Roldanquit (Quit: Samuel_Roldan)
22:39:15  * tilgoviquit (Quit: No Ping reply in 180 seconds.)
22:40:47  * tilgovijoined
23:08:43  * marfarmaquit (Ping timeout: 276 seconds)