Wednesday, 2010-08-25

*** ArneBab is now known as ArneBab_away00:15
*** akila87 has joined #cc00:19
*** mralex has quit IRC00:24
*** JED3 has quit IRC00:26
*** akozak has quit IRC00:38
CC_HudsonProject cc.license build (18): STILL FAILING in 2 min 6 sec:
CC_Hudson* Christopher Allan Webber: Use string.template for doing mapping, because apparently mapping in00:50
CC_Hudson* Christopher Allan Webber: Most of the expected formatter output added, though not used in tests yet00:50
CC_HudsonProject deedscraper build (11): STILL FAILING in 1 min 35 sec:
CC_HudsonProject cc.engine build (21): STILL FAILING in 2 min 11 sec:
CC_HudsonProject metadata_scraper build (13): STILL FAILING in 58 sec:
CC_HudsonProject cc.license build (19): STILL FAILING in 1 min 10 sec:
CC_HudsonChristopher Allan Webber: Moving accidental quotes and parens01:04
*** nyergler has quit IRC01:15
*** paulproteus has quit IRC01:57
*** greg-g has quit IRC01:57
*** dithyramble has quit IRC01:57
*** pyrak has quit IRC01:59
*** akila87 has left #cc03:23
*** erlehmann has joined #cc03:31
*** ArneBab_away is now known as ArneBab03:42
*** MarkDudeva has joined #cc04:41
*** MarkDudeva is now known as MarkDude04:41
*** MarkDude has joined #cc04:42
*** MarkDude has quit IRC04:50
*** MarkDude has joined #cc04:54
*** pmiller has joined #cc05:12
*** Andre_Re1 has joined #cc05:29
Andre_Re1Hello there, just a quick question05:29
Andre_Re1Does CC-BY-SA actually mean, that Ihave to redistribute a derivated work under CC-BY-SA or can I also redistribute the derivate as CC-BY?05:31
*** Odemia has joined #cc05:32
*** erlehmann has quit IRC05:49
*** pmiller has quit IRC05:51
*** Odemia has quit IRC05:52
*** JoiIto has quit IRC06:04
*** Kaetemi has joined #cc06:40
*** tantek has quit IRC06:44
*** MarkDude has quit IRC06:58
*** thisisparker has quit IRC07:07
*** MarkDude has joined #cc07:09
*** tantek has joined #cc07:16
*** Andre_Re1 has left #cc07:27
*** pmiller has joined #cc07:53
*** thisisparker has joined #cc08:02
*** tantek has quit IRC08:08
*** tantek has joined #cc08:08
*** bassel has quit IRC08:14
*** Odemia has joined #cc08:58
*** Odemia has quit IRC09:15
*** Odemia has joined #cc09:19
*** akila87 has joined #cc09:27
*** tantek has quit IRC09:35
*** bassel has joined #cc09:51
*** pmiller has left #cc09:56
*** tantek has joined #cc10:26
*** tantek has joined #cc10:41
*** akila87 has left #cc10:43
*** tantek has quit IRC10:52
*** tantek has joined #cc10:54
*** akila87 has joined #cc11:20
*** Odemia has quit IRC11:26
*** karlcow has quit IRC11:46
*** karlcow has joined #cc11:55
*** bassel has quit IRC12:07
*** oshani has joined #cc12:12
*** tantek_ has joined #cc12:31
*** tantek has quit IRC12:31
*** tantek_ is now known as tantek12:31
*** akila87 has left #cc12:38
*** akila87 has joined #cc12:50
*** tantek has quit IRC13:00
*** EzOne has joined #cc13:11
*** EzOne has left #cc13:11
*** akila87 has quit IRC13:38
*** paulproteus has joined #cc14:02
*** paulproteus_ has joined #cc14:08
*** dithyramble has joined #cc14:08
*** pyrak has joined #cc14:08
*** paulproteus has quit IRC14:12
*** tantek has joined #cc14:13
*** tantek has quit IRC14:15
*** greg-g has joined #cc14:17
*** nkinkade has joined #cc14:40
alsk___anttil, i think SA means it's CC-BY-SA ?14:46
paroneayeankinkade: is being really slow or something?15:06
paroneayeaI keep notice that on it seems to be hanging on loading images from there15:07
nkinkadeparoneayea: a8 has been a little slow lately, and I'm not sure what to attribute it to.15:08
nkinkadeIn my mind it has coincided with the the staging branch of the API being moved to dev.15:09
*** tantek has joined #cc15:14
paroneayeajust saw the nagios alerts, looks like are down15:16
paroneayeankinkade: wtf?15:18
paroneayea3 ms15:18
paroneayeascared myself for a minute there15:19
paroneayeathought we got hijacked15:19
nkinkadeI got the message that /apps/deed was down earlier this morning.15:21
nkinkadeHonestly, I'd like to move to it's own server.15:21
nkinkadeI don't trust the API anymore ... even if the API isn't the problem.15:21
nkinkadeI experienced those intermittent problems with the API timing out when viewing the /international/xx/ pages.  That still hasn't been explained.15:22
*** marucs has quit IRC15:22
paroneayeayeah, I also see calls to slow at times too15:23
*** SiDi has joined #cc15:42
SiDiHey there. Does anyone know where i may find a docbook version of cc-by-sa-3.0 please?15:42
*** akila87 has joined #cc15:42
greg-gparoneayea: feature request ^^^ "add ability to export license legal code in PDF, LaTeX, Plain Text, and Docbook."  :P15:45
paroneayeagreg-g: it does seem to be a common request15:46
paroneayeapart of the problem is that the legalcode HTML is not generated by the engine15:46
greg-goh :/15:46
paroneayeait's hand-written by our translators15:46
paroneayeaso that the html version of the legalcode is canonical and never changes15:46
greg-goh well15:47
greg-gthat would be a big understanding then, no?15:47
paroneayeawe could probably create PDF and plaintext of the core 6 licenses + cc0 maybe15:47
paroneayeaall the ported ones though I think that's unlikely to ever happen15:48
SiDiparoneayea: i think it'd be great to have them maintained under a format from which it can be exported15:49
SiDiwith docbook you'd get PDF/LaTeX/HTML under hand already15:49
SiDicause now, i'm so lazy to do the CC 3.0 DocBook by myself that i'll license my doc under the 2.5 instead :D15:49
SiDi(i suppose CC bysa 2.5 is not compatible with GNU FDL?)15:49
greg-gif by compatible you mean interoperable, you are correct15:51
SiDigreg-g: so i could switch from cc 2.5 to FDL?15:56
SiDi(see how lazy i am:  i choose a license based on whether there is a .xml docbook file available)15:57
*** akila87 has left #cc16:09
*** tantek has quit IRC16:10
greg-gSiDi: you *could* but I wouldn't recommend it :/16:13
SiDigreg-g: why?16:13
greg-gbecause more and more work is being produced under CC:BY-SA instead of the FDL16:14
paroneayeaSiDi: if you choose BY-SA, your work will be compatible with a greater number of works16:15
paroneayeaie, stuff on wikipedia, etc16:15
*** JoiIto has joined #cc16:26
*** akila87 has joined #cc16:32
*** thisisparker has quit IRC16:37
*** tantek has joined #cc16:39
*** tantek has quit IRC16:42
*** MarkDude has quit IRC16:42
*** JED3 has joined #cc16:46
*** Endiannes has joined #cc17:02
*** karlushi has quit IRC17:02
EndiannesI'm using a .mp3 file licences with cc 2.0, am I allowed to convert it to .ogg without modifying anything else?17:02
*** papy|away is now known as papyromancer17:05
*** JED3 has quit IRC17:06
nkinkadeparoneayea: Hey ... is the machinery in place for commits to to show up automatically on staging?17:17
nkinkadeThat is, is it all up and working to your knowledge?17:18
paroneayeankinkade: no :)17:18
paroneayeait automatically makes the tarball17:18
paroneayeabut doesn't do a buildout and restart17:18
nkinkadeparoneayea: What is left to be done for that to work?17:18
nkinkadeI think that is something we really need to address sooner than later now that we have officially moved to Transifex.net17:19
paroneayeankinkade: not much.  I could add to the cronjob so that it buildouts and restarts apache after it makes the egg release17:19
nkinkadeparoneayea: Was that how we decided to tackle it, with a cronjob rather that post-receive?17:19
nkinkadeJust trying to remember.17:19
paroneayeawell, it seems sensible to use the cronjob now17:19
paroneayeasince we're making the egg via cronjob.17:20
nkinkadeparoneayea: How often does it run?17:20
paroneayeaevery 15 mins17:20
paroneayeawould that be sufficient?17:20
nkinkadeparoneayea: Every 15 mins seems fine to me.17:20
nkinkadeIdeally it would all happen immediately via post-receive.17:20
nkinkadeIs John on vacation?17:21
paroneayeano he's running the conference call currently being run17:21
paroneayeaI think.17:22
*** mralex has joined #cc17:27
nkinkadeOoops.  I'm not on that call.17:28
*** JED3 has joined #cc17:29
nkinkadeparoneayea: how much time would it take for you to finish the cron job to have it update staging?17:30
nkinkadeIs it really just a couple lines in a script?17:31
paroneayeankinkade: yeah, just a couple of lines17:31
nkinkadeparoneayea: Would you be willing to finish that today? :-)17:31
paroneayeankinkade: sure :)17:31
nkinkadeI'll write to CC Sweden to tell them that it should be updating by later today, then.17:31
*** notb has joined #cc17:38
nkinkadeJED3: It seems that /apps/deed and /apps/triples are down ... have been since earlier today.17:38
*** alsk___ has quit IRC17:40
*** notb_ has joined #cc17:43
JED3nkinkade: thank you, how do i change the permissions set by the log roller?17:43
nkinkadeJED3: In /etc/logrotate.d/metadata-scraper17:44
*** notb has quit IRC17:45
JED3nkinkade: perfect thank you17:45
paroneayeankinkade: done17:51
nkinkadeparoneayea: Thanks!17:51
*** Endiannes has quit IRC18:01
*** alsk___ has joined #cc18:05
*** notb_ has quit IRC18:08
*** thisisparker has joined #cc18:14
*** SiDi has quit IRC18:35
*** MarkDude has joined #cc18:48
*** alsk___ has quit IRC18:51
paroneayeaJED3: do you know if there's a way to tell buildout, "Just update this one package"?18:54
JED3paroneayea: rm the egg and buildout, thats my solution normally18:54
JED3and give the non-newest flag so the others arent updated18:55
JED3-N i believe18:55
paroneayeaJED3: maybe I should do that so as to save the server the extra load of running buildout every 15 minutes18:55
*** MarkDude has quit IRC18:55
paroneayeaJED3: oh, will that work with tarballs?18:57
paroneayeaor do you have to untar-and-do-something18:57
*** erlehmann has joined #cc18:58
CC_HudsonProject cc.i18n build (41): SUCCESS in 37 sec:
CC_HudsonTransifex robot: l10n: Updated Georgian (ka) translation to 3%19:22
*** ArneBab is now known as ArneBab_away19:29
*** click170 has quit IRC19:32
*** MarkDude has joined #cc19:33
*** akila87 has quit IRC19:49
*** tantek has joined #cc19:56
CC_HudsonProject cc.license build (20): STILL FAILING in 1 min 55 sec:
CC_Hudson* Christopher Allan Webber: String test for pdmark worktitle in place20:05
CC_Hudson* Christopher Allan Webber: Tests for most of these pdmark formatter possibilities, but test_cc0 failing20:05
CC_Hudson* Christopher Allan Webber: Forgot to switch alt text from Public Domain Mark -> CC0 on that alt tag20:05
CC_Hudson* Christopher Allan Webber: We don't put the dct namespace in the <p> tag when we don't use it20:05
JED3paroneayea: sorry went ark20:06
paroneayeaJED3: np20:06
JED3will it work with tarballs? yes it should20:06
paroneayeaJED3: didn't seem to, I tested :\20:07
JED3hmm, can you describe what you are trying to do?20:07
JED3updating an cc.i18n dependency?20:07
paroneayeacp latest cc.i18n*.tar.gz from basket/ over to eggs/20:08
paroneayeathat way I don't have to run ./bin/buildout20:08
JED3ohh okay20:08
paroneayeawhich seems to work just fine20:08
paroneayeaI just suspect it'll eat up more CPU than necessary20:08
JED3really that works? won't you start getting version conflicts?20:08
*** tantek_ has joined #cc20:09
paroneayeaJED3: do you have a better solution? :(20:09
paroneayeayeah, it's fragile.  Things could blow up eventually and bring down staging20:10
paroneayeabut using buildout I don't know of a solution20:10
JED3whats wrong with...20:10
JED3rm -fr eggs/cc.i18n....egg/ ; bin/buildout -N ?20:10
paroneayeaJED3: it won't do anything with an egg that's not untarred and develop'ed20:11
paroneayeawhich you can't do manually with buildout20:11
JED3paroneayea: what checkout is this for?20:12
paroneayeaJED3: sanity on staging20:12
*** tantek has quit IRC20:12
JED3paroneayea: setuptools with fetch the egg from basket and untar itself won't it?20:13
paroneayeaJED3: yeah!  If you don't do -N :)20:13
paroneayeaoic what you mean20:13
paroneayeathat would work20:13
*** tantek_ has quit IRC20:13
JED3are you sure? if you delete the egg first it should work20:13
paroneayeayeah I'll try that20:13
paroneayeaslight chance of it erroring out the server20:14
paroneayeain case some file is not cached or something weird20:14
paroneayeadon't think that would happen though20:14
paroneayea(in the brief period when the egg is gone)20:14
JED3yeah, thats true, you could always sleep() in the shell script20:15
*** thisisparker has quit IRC20:17
CC_HudsonProject cc.license build (21): STILL FAILING in 1 min 17 sec:
CC_Hudson* Christopher Allan Webber: Escapefest test for pdmark formatter20:19
CC_Hudson* Christopher Allan Webber: Removed some unnecessary whitespace20:19
*** tantek has joined #cc20:19
*** tantek has quit IRC20:24
paroneayeankinkade: I forget, is nyergler doing things today?20:27
paroneayeathings that are incompatible with a weekly tech check-in20:27
nkinkadeparoneayea: Not sure.  I haven't seen him online today.20:27
paroneayeaJED3: you know?20:27
mralexhe's in DC20:27
nkinkadeThat would explain it.20:27
JED3paroneayea: yeah i dont think we'll be having our call20:29
*** MarkDude has quit IRC20:29
*** tantek has joined #cc20:30
*** oshani has quit IRC20:34
*** tantek has quit IRC20:35
*** thisisparker has joined #cc20:35
*** erlehmann has quit IRC20:39
*** MarkDude has joined #cc20:40
*** tantek has joined #cc20:41
*** tantek has quit IRC20:45
*** tantek has joined #cc20:51
*** thisisparker has quit IRC20:54
*** tantek has quit IRC20:56
*** thisisparker has joined #cc21:00
*** tantek has joined #cc21:02
*** MarkDude has quit IRC21:04
*** tantek has quit IRC21:06
nkinkadeJED3: This URL seems to be loading quite slowly.  My last run was 10s.  On one run a minute ago it timed out.21:07
nkinkadeI just have this impression that the API has got really slow lately ... mostly just anecdotal evidence.21:08
JED3wow really, there have been some serious fluctuations in the api performance21:08
*** thisisparker has quit IRC21:08
nkinkadeYeah, I'm wondering what it could all be about.  Maybe it's something else causing the API to be slow?21:08
JED3hmm, you know one thing that isn't there that should be, is varnish caching some of the calls that are basically static21:09
JED3this jurisidictions.js call is one21:09
JED3which gets clobbered by requests from cc.org21:10
nkinkadeJED3: Yeah.  I'm really noticing slowdowns on just about any /international/xx/ page.21:11
nkinkadeEvery page load there requests jurisdictions.js.  In fact, I think every page load at CC(!) makes a call to that script.21:11
JED3hmm yeah...i wonder if i should use nry's cache middleware21:12
JED3or we could configure varnish to handle it21:12
JED3which is probably easier21:12
JED3its a8 thats slow for me21:12
JED3just opening the docs for the api, is about to time out on me21:13
*** tantek has joined #cc21:13
JED3which isn't served by the api process21:13
nkinkadeJED3: But if you look at resource usage on a8, things look okay.21:13
JED3 nkinkade does this load quickly for you?21:13
nkinkadeYeah, and this morning images were coming back slow for
JED3yeah, i know, the usages look fine21:14
nkinkadeJED3: It took nearly 10s to load doctool.js21:14
JED3nkinkade: does this load fine for you?21:14
JED3it just timed out on me21:14
nkinkadeJED3: Yeah it loaded for me, but it took nearly 10s.21:15
nkinkadeJED3: the requests *do* hit varnish, but they don't get cached because the requests come with cookies.21:15
JED3oh really?21:16
nkinkadeAnd Varnish doesn't cache requests that contain cookies ... at least not by default.21:16
nkinkadeHowever, doesn't set any cookies that I know of, or that my browser tells me about.21:16
nkinkadeIt sends them because of cookies set via
nkinkadeJED3: Wouldn't it be safe to pretty much cache every single response from the API?21:17
JED3are you sure varnish is caching the rest/dev|staing/1.5/etc requests?21:17
nkinkadeI mean, things change in the API at a glacial pace.21:17
*** tantek has quit IRC21:17
nkinkadeJED3: Varnish is not caching that stuff, though the request does pass through Varnish.21:17
nkinkadeAnd in any case, Varnish's default timeout is 180s, I think.  And no data from the API is going to change even every 3 months, much less every 3 minutes.21:18
nkinkadeJED3: Should I tell Varnish to start caching all API requests?21:18
JED3nkinkade: thats true, but some of the api calls aren't really cacheable due to allowed parameters21:19
nkinkadeJED3:  Hmm.  So Varnish won't take into consideration the query string?21:19
nkinkadeThat could be.21:19
nkinkadeI need to check.21:19
JED3nkinkade: well even if it did it might cause some cache bloat21:20
JED3parameterized requests are unlikely to be called numerous times by different clients21:21
nkinkadeHow many different unique URLs are hit on any given day at the API?21:21
nkinkadeJED3: Right, but I think the largest consumer of the API are other CC sites, right?21:21
nkinkadeLike the license chooser and the perhaps the deeds.21:21
JED3nkinkade: yes, the jurisdiction calls should definitely be cached21:22
nkinkadeJED3: Who uses the API except for CC?  At least that we know of?21:22
*** tantek has joined #cc21:23
JED3well the wplicense does, jswidget, a few other platforms21:24
JED3but probably 95% us21:24
nkinkadeJED3: I was just looking at the API Apache logs and it seems that there have been around 60,000 requests today, from around 6,500 unique IP address, only 3766 of which requests were from a5.  That sounds odd to me.21:27
nkinkadeThat would indicate that a mere 5% of the requests are coming from a5.21:27
nkinkadeAt least for today.21:27
JED3wow, thats strange, because i'm awking the logs right now all i'm seeing are from a521:28
*** tantek has quit IRC21:28
JED3there must be a burst in there somewhere then21:28
JED3because right now, its all a521:28
*** thisisparker has joined #cc21:28
*** tantek has joined #cc21:34
JED3nkinkade: yeah i think we're our own worst enemy here, are you sure those metrics you mentioned were accurate?21:37
JED3look at the file "api-clients.log" in my home dir on a821:38
*** tantek has quit IRC21:39
*** JoiIto has quit IRC21:40
nkinkadeJED3: I was looking at a5:/var/log/apache/
nkinkadeIs that the wrong log?21:41
*** thisisparker has quit IRC21:41
nkinkadeWhat does that log log? :-)21:41
JED3nkinkade: yes, api is a8 :)21:42
JED3nkinkade: beats me!21:42
nkinkadeJED3: Right, I meant a8:21:43
nkinkadeJED3: Where did api-clients.log come from?21:44
JED3nkinkade: this...21:44
JED3sudo cat /var/log/apache2/ | awk '{print $11}' | grep -v "" > ~/api-clients.log21:44
*** tantek has joined #cc21:45
JED3thats only looking at the referrers though21:45
*** MarkDude has joined #cc21:48
nkinkadeJED3: Sorry, I was looking at the source address, not the referrer.21:49
nkinkadeFor some reason I felt that the source would be a5.21:49
JED3nkinkade: i just updated that log file with source ip & referer21:49
nkinkadeI thought the API calls happened behind the scenes.21:49
*** tantek has quit IRC21:50
nkinkadeThat is like User -> a5 -> api -> a5 -> User21:50
nkinkadeOr something vaguely like that.21:50
nkinkadeIs it purely javascript that accesses the API, then?21:50
JED3yes, its client side21:51
JED3its <script src=""> which is requested by a users browser21:51
*** tantek has joined #cc21:55
nkinkadeJED3: So, out of some 60,000 requests, there were only 494 unique request URLs.21:58
nkinkadeI'm not really worried about cache bloat on the side of Varnish over ~500 distinct URIs.  Beyond that, most of those requests return tiny amounts of text, which don't take up much space.21:59
*** tantek has quit IRC22:00
JED3nkinkade: very true, lets cache it all then22:00
nkinkadeJED3: I'll turn on caching via Varnish right now, and let's just see if we notice any difference right away ... more snappiness.22:01
JED3nkinkade: ok22:01
nkinkadeJED3: My god ... jurisdictions.js is now loading in .117ms22:06
*** tantek has joined #cc22:06
JED3nkinkade: woot!22:06
nkinkadeAnd /docs/ is loading ridiculously fast too.22:06
nkinkade... at least for me.22:06
JED3nkinkade: awesome, me too22:06
JED3i didnt realize we werent already caching the js related calls22:07
nkinkadeJED3: I think we just sped up page loads for lots of CC pages by a factor of 2 or 3.22:21
CC_HudsonProject cc.i18n build (42): SUCCESS in 32 sec:
CC_HudsonTransifex robot: l10n: Updated Russian (ru) translation to 99%22:21
JED3wow good point, this uses the api server side doesnt it?22:22
nkinkadeThat URL used to consistently take many seconds to load, sometimes up to 10s.  Now after the initial load and subsequent refreshes it load in about 1s.22:22
JED3via a php script or something right?22:22
nkinkadeJED3: It hits the API twice .. once for the jurisdiction list in the header behind the "International" link, and again behind the scenes to fetch that list of current licenses.22:22
nkinkadeThe one fetched by the new API call you created.22:23
JED3nkinkade: must snappier i see22:23
JED3nkinkade: ahh, will varnish cache resources identified by the query strings args?22:26
*** Kaetemi has quit IRC22:27
*** Kaetemi has joined #cc22:28
nkinkadeJED3: I believe so.22:30
JED3nkinkade: and varnish will cache requests for get and post?22:30
nkinkadeI turned caching off for now.22:30
nkinkadeJED3: I presume so.22:30
nkinkadeThough we can make that distinction in the Varnish config file.22:31
nkinkadeThat is cache this pattern, but not that pattern ... cache GET, but not POST, etc.22:31
JED3nkinkade: okay, lets doublecheck as to whether varnish uses q params in the cache's keys22:32
JED3nkinkade: looks like get parameters are fine22:34
nkinkadeJED3: How/where are you checking this?22:34
JED3so if thats the case, only /rest/(dev|staging|1.5)/license/(standard|publicdomain|zero|recombo)/issue will need to be excluded22:36
JED3nkinkade: looking at the varnish docs22:36
JED33rd question, makes me believe that varnish keeps the query string by default22:37
nkinkadeJED3: What goes to the API via those requests you mention that is user specific?22:39
nkinkadeI thought all that stuff was handled via cc.engine.22:39
nkinkadeDoes cc.engine just make API calls?22:39
JED3nkinkade: the engine doesn't use the api at all, except for the jurisdictions.js in the templates22:40
nkinkadeSo what requests are user-specific?22:40
JED3nkinkade: as in which ones are highly parameterized?22:41
nkinkadeJED3:  parameters don't worry me, just ones that return user-specific data.22:41
JED3just those two22:42
nkinkadeJED3: Then what is an example URI to the API that would break functionality if it were cached?22:42
JED3well if /issue was cached then that would be an example22:43
JED3that url is posted to with a set of user parameters22:44
nkinkadeJED3: What might a full URL be that would break the system?22:45
JED3so just issue actually needs to be exempt22:45
mralexJED3: When the song picker list for Populist is long (as it is right now), it takes forever to display the popup and register a request - chrome timed out a few times22:46
nkinkadeJED3: The above URL just returns "None" for me.22:47
nkinkadeI guess I'm missing the params.22:47
nkinkadeOh, it's a POST.22:47
nkinkadeI think I must be being dense here, because it's not clear to how the results of that URL will differ from one person to the next, as long as Varnish respects all query paramaters.22:48
*** mralex has quit IRC22:50
*** JED3 has quit IRC22:50
CC_HudsonProject cc.i18n build (43): SUCCESS in 28 sec:
CC_HudsonTransifex robot: l10n: Updated Russian (ru) translation to 100%22:51
*** mralex has joined #cc22:55
*** JED3 has joined #cc23:02
nkinkadeJED3: To be sure, as I'm writing the regex, for the URL portion, this would match all requests for /issue that should not be cached "/issue$"23:07
nkinkadeThat is, there will never be anything, even a slash, after /issue, right?23:08
JED3nkinkade: nope that'd return 40423:09
nkinkadeJED3: In that case it seems as if we are safe to skip caching any requests to the API of the form /issue$ or /get$23:10
nkinkadeDoes that sound about right to you?23:10
JED3nkinkade: yeah, i actually think we can cache get, but better safe than sorry, dont cache issue or get23:11
nkinkadeJED3: Out of ignorance on my part, how does /get differ from /issue, except that it take query params instead of POST variables?23:12
JED3nkinkade: thats the only different, the request method23:12
JED3get is GET, issue is POST23:13
JED3no redirection between the two either23:13
nkinkadeJED3: So why would you think that the /get call could be cached, and /issue not?23:14
nkinkadeJust because /get isn't really used by anyone?23:14
JED3no because all of the parameters to /get are in the query string23:15
JED3which is kept in the varnish cache key23:16
nkinkadeJED3: That makes sense, you're right.23:40
nkinkadeBut I guess it doesn't really make sense to cache single use URLs anyway, so I just excluded /get? as well.23:40
JED3nkinkade: awesome, is varnish enabled right now?23:41
nkinkadeJED3: Yes.  I re-enabled caching in Varnish.23:41
nkinkadeAnd I verified it isn't caching for /issue and /get?23:42
nkinkadeSeems to be caching for everything else.23:42
nkinkadeVarnish is really great.23:42
JED3nkinkade: great!23:42
JED3yes it is.23:42
nkinkadeI can't imagine using Squid for anything these days.23:42
JED3nkinkade: why is that?23:43
nkinkadeBecause I don't know of anything that Squid does that Varnish doesn't improve upon.23:44
nkinkadeThough I'm sure someone could find numerous things.23:44
nkinkadeIt's just for my/our use cases Varnish absolutely simplifies things.23:44
nkinkadeThe config is so easy.23:44
JED3nkinkade: ahh yeah, the conf examples on the squid site look really complicated23:47

Generated by 2.6 by Marius Gedminas - find it at!