[00:14] if there's a review of https://codereview.appspot.com/8811043/ we could get it in under the wire [00:14] which would be nice :-) [00:15] and it is very small [00:16] * davecheney looks [00:22] gary_poster: this diff is dirty [00:22] there is some bzr .THIS crap in thre [00:22] davecheney, I was assuming that it was a clean up [00:23] davecheney, might a repush clean it up? [00:23] Makyo, ^^^ could you try a repush? [00:23] davecheney, than kyou for looking [00:23] actually, this removes a mistaken checkin [00:23] let me publish my review, there is some more dirtyness [00:23] ok thanks [00:24] hazmat: this was your fault! [00:25] done [00:25] thumper: https://codereview.appspot.com/8811043/ [00:25] can you give the gui guys a second LGTM [00:25] gary_poster: or you could just commit it [00:25] if time is a factor [00:25] davecheney, thanks much. I'll check what is possible, [00:25] * thumper looks [00:27] done [00:28] * thumper off to the gym shorly for some sparring [00:28] gary_poster: Makyo you've got two, make those changes and fire at will [00:28] awesome thanks very much davecheney and thumper === thumper is now known as thumper-afk [00:38] does anyone have a link to last week's hangout agenda ? [00:38] m_3: ping [00:46] okay, have the other bug down as well [00:49] davecheney: hey [00:49] mgz: we're stuc [00:49] k [00:49] m_3: how's it going ? [00:49] mgz: I'm leaving it be atm [00:50] davecheney: seems hung up on something [00:50] hang on, changing hosts [00:50] m_3: tell me where it hurts [00:51] davecheney: we brought up 200 earlier to checck on adding incremental nodes via nova cli [00:51] m_3: I saw it's not increasing, but didn't see anything in the log [00:51] davecheney: mgz played a bit [00:51] davecheney: then added another 200 (juju add-unit hadoop-slave -n200) [00:51] I have two of the three issues I've seen tracked down [00:51] davecheney: it added 15 more and hung [00:51] * m_3 gotta relocate... getting kicked out of the expo area [00:52] mgz: i've seen an issue where the PA will get into a loop if there are problems creating security groups [00:52] back on after while... I'll leave it up for y'all [00:52] m_3: mgz: i suggest destroying this instance [00:52] bzr pull to rev 1164 [00:52] then trying again [00:53] 2013/04/16 23:41:06 ERROR provisioning worker: failed to GET object provider-state from container goscale2-1 [00:53] caused by: https://region-a.geo-1.objects.hpcloudsvc.com/v1/17031369947864/goscale2-1/provider-state%!(EXTRA string=failed executing the request) [00:53] i think i have a fix for these formatting errors [00:55] mgz: i'd like to destroy this envbironment and start again with a fresh build [00:56] okay, we could probably do with fixing a couple of the dumb issues too [00:56] rev 1164 cuts down on 66% of the log span [00:56] spam [00:56] that particular one is probably api flood issue again [00:56] i'll grap the all machines log [00:57] the blow away the environment [00:59] ubuntu@juju-hpgoctrl2-machine-0:~$ scp 15.185.160.145:/var/log/juju/all-machines.log . [00:59] all-machines.log 20% 141MB 5.6MB/s 01:38 ETA [00:59] that is a great network hp [01:03] filed bug 1169778 [01:03] davecheney: compress first :) [01:04] mgz: sure [01:04] but 5.4Mb/s is shit [01:04] my pandaboard can do better than that [01:04] mgz: it's nice to know you can resume juju destroy-environment [01:08] mgz: m_3 we have 7 stray nova instances after destroy-environemnt [01:08] hmm [01:08] actually only 2 [01:08] never mind [01:12] one was mine, that I killed, and one is the manage one? [01:12] fixed it [01:12] wc -l was the wrong thing to tuse [01:15] mgz: m_3 deploying 200 hadoop-slave units [01:17] 100% of the time is being consumed waiting on hp to respond to new instance requests [01:57] davecheney: hey [01:57] davecheney: how's it going? [01:58] davecheney: btw, you never want to use 'hadoop-slave' the charm... (long story) [01:58] m_3: 2013/04/17 01:52:25 NOTICE worker/provisioner: started machine 200 as instance 1509985 [01:58] davecheney: make sure you're doing `juju deploy hadoop hadoop-slave -n200` [01:58] m_3: sorry, i was looking in the history, that is all I could find [01:58] ack [01:58] nm, we can destroy the instance [01:58] always ran from bin/hadoop-stack [01:59] ok, let me nuke this environment [01:59] (it worked fine btw [01:59] rev 1164 is much less chatty [01:59] davecheney: did it get above 200? [01:59] davecheney: we have perms to go to 2000 [01:59] just did -n200 [01:59] were just having probs right near the 200 mark [01:59] i'll bootstrap a fresh env [01:59] do bin/hadoop-stack [01:59] I did a `-n1997` before [02:00] yeep! [02:00] thati'll take hours [02:00] did you get mgz's changes in? [02:00] no, but the logging change went in [02:00] yup... about 8hrs or so [02:00] so we can actually see what is going on [02:00] ah, cool [02:00] wanna do 300? [02:00] ok [02:00] or shall we go big [02:00] 500 should see us throught to drinks [02:00] how is ODS ? [02:00] is the marketing completely off tap [02:01] ? [02:01] dude [02:01] hookers and blow ? [02:01] it's pretty over the top [02:01] haha [02:01] yeah, all hookers and blow [02:01] supposed to meet kapil in the lobby for dinner in a bit [02:01] actaully, card tables, laptops and tshirts [02:02] yeah :( [02:02] better than couch, popcorn, and underwear at least [02:02] m_3: i don't need to go to a conference for that [02:04] m_3: we're good to go [02:04] hit it with the old -n300 [02:05] hmm, that isn't good [02:05] let me drive for a second [02:05] ok, still piss farting around [02:05] the last bootstrap was much faster [02:06] ack [02:06] still scroblling in apt [02:06] ├─cloud-init-cfg─┬─add-apt-reposit [02:06] │ └─sh───tee [02:06] wow [02:06] that is slow [02:06] this instance may fail to bootstrap [02:07] might just killl it [02:07] do it [02:07] oh wait [02:07] yeah, that is really slow [02:07] nuke it [02:08] one that one's up just run bin/hadoop-stack [02:08] I'll chcek in after dinner if that's cool with you? [02:08] m_3: ok, is that configured for -n300 ? [02:08] maybe you can catch what's going wrong near the 200 mark [02:08] totolly cool [02:08] yup [02:08] you go have dinner with K [02:08] awesome thanks man... ttyl [02:09] oh, and let's archive the logs in ~/arch/ [02:09] sure, i've been scp'ing them onto my machine [02:09] will fix [02:09] at least to keep some sort of record of what we're doing [02:09] thanks [02:09] actually, i'll put it in google drive [02:09] I'm still rsyncing the whole control node's /home/ubuntu offsite daily [02:09] perfect [02:10] probably isn't a bad idea [02:10] make sure you use rsync -z [02:10] (-azvP) [02:10] cool... /me food [02:10] thanks man [02:12] m_3: also, status now shows relations [02:13] 2013/04/17 02:13:05 DEBUG environs/openstack: openstack user data; 2708 bytes [02:13] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/6" [02:13] \o/ [02:13] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/6" [02:13] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/5" [02:13] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/5" [02:13] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/6" [02:13] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/6" [02:14] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/5" [02:14] 2013/04/17 02:13:05 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/5" [02:14] if this is an error, why does the provisioner not stop [02:14] davecheney: that doesn't look good [02:14] ? [02:15] 2013/04/17 02:12:29 INFO environs/openstack: starting machine 14 in "goscale2" running tools version "1.9.14-precise-amd64" from "https://region-a.geo-1.objects.hpcloudsvc.com/v1/17031369947864/goscale2-1/tools/juju-1.9.14-precise-amd64.tgz" [02:15] 2013/04/17 02:12:29 DEBUG environs/openstack: openstack user data; 2710 bytes [02:15] 2013/04/17 02:12:40 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/1" [02:15] 2013/04/17 02:12:40 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/1" [02:15] 2013/04/17 02:12:40 ERROR settings for entity with unrecognized key "r#0#provider#hadoop-master/0" [02:15] 2013/04/17 02:12:40 ERROR settings for entity with unrecognized key "r#1#provider#hadoop-master/0" [02:15] 2013/04/17 02:12:40 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/1" [02:15] 2013/04/17 02:12:40 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/1" [02:15] 2013/04/17 02:12:40 ERROR settings for entity with unrecognized key "r#0#provider#hadoop-master/0" [02:15] 2013/04/17 02 [02:15] starts to screw up here [02:17] caused by: https://az-2.region-a.geo-1.compute.hpcloudsvc.com/v1.1/17031369947864/servers%!(EXTRA string=failed executing the request) [02:17] i really want to fix this stupid formatting errror [02:47] shit shit shit [02:47] the 1.9.14 release cannot seem to download any tools [02:47] however when I build the release from source [02:47] it can ! [02:51] 2013/04/17 02:50:21 DEBUG checking tools 1.9.14-precise-amd64 [02:51] 2013/04/17 02:50:21 INFO environs/openstack: starting machine 298 in "goscale2" running tools version "1.9.14-precise-amd64" from "https://region-a.geo-1.objects.hpcloudsvc.com/v1/17031369947864/goscale2-1/tools/juju-1.9.14-precise-amd64.tgz" [02:51] 2013/04/17 02:50:21 ERROR worker/provisioner: cannot start instance for machine "298": cannot find image satisfying constraints: failed to get list of flavours [02:51] caused by: https://az-2.region-a.geo-1.compute.hpcloudsvc.com/v1.1/17031369947864/flavors%!(EXTRA string=failed executing the request) [02:51] caused by: Get https://az-2.region-a.geo-1.compute.hpcloudsvc.com/v1.1/17031369947864/flavors: lookup az-2.region-a.geo-1.compute.hpcloudsvc.com: Temporary failure in name resolution [03:12] i'm having a lot of trouble validating the 1.9.14 release [03:13] is anyone in a position to install 1.9.14 from ppa:juju/devel [03:13] and try [03:13] (don't forget to make sure you're calling /usr/bin/juju, not ~/bin/juju) [03:33] hazmat, if you're there, I don't think we necessarily want to destroy machines [03:34] hazmat, I suspect the deploy-ubuntu-to-prime, force-machine-to-deploy (anti-)pattern will be popularly used [03:54] davecheney, thumper: if you're there, can I get a trivial on https://codereview.appspot.com/8815043 please? [03:57] fwereade: i'll take a look [03:57] fwereade: i'm very concerned that 1.9.14 is stillborn [03:58] i'm thinking of proposing we move to daily point releases [03:58] so m_3 and I stop using the development chain [03:58] davecheney, I don't think it'll build, given this stuff [03:58] davecheney, it's an old test file that somehow crept in [03:59] wut ? [04:00] davecheney, at a casual glance it looks like it slipped in from the force-machine branch [04:00] this is the second turd to fall out of that branch [04:00] davecheney, although it's not quite clear how [04:00] without being rude, 4 people looked at that branch [04:00] davecheney, ouch :( [04:00] and nobody spotted the .THIS file there [04:02] fwereade: LGTM, please submit [04:03] davecheney, happening right now [04:03] davecheney, looks like that file slipped in somewhere between 1 and 5 days ago after the main round of reviewing [04:04] so that means the submitter did not run the tests before submitting [04:04] do we need to make another adding to .lbox.check ? [04:04] /s/adding/addition [04:05] I never found out what happened to the bot proposal but that also seems to be stillborn [04:05] in its absence, maybe we do [04:05] hi fwereade [04:06] what are you doing up? [04:06] thumper, fell asleep early last night, back up now to merge a whole pile of things and do some reviews [04:06] fwereade: we can't have a bot til the tests pass reliably [04:06] that was my understanding [04:06] fwereade: oh, I have four reviews up? [04:06] * thumper goes to drink the coffee [04:08] davecheney, if we can't have a bot without reliable tests but we can't commit reliably without a bot *something* needs to change [04:09] fwereade: i think we can at least do a test compile of the tests [04:09] that would ahve caught this error [04:09] but it does seem a bit futile [04:09] we're like the TSA [04:09] running after the horse [04:09] davecheney, oh! how do we compile tests without running them? [04:10] davecheney, actually, don't tell me, I might start doing it ;p [04:11] fwereade: i'm sure there is a way [04:11] (cd /tmp ; go test -c $PKG) will do it [04:12] davecheney, emphatic +1 then [04:17] fwereade: http://paste.ubuntu.com/5714987/ [04:17] can anyone else test the 1.9.14 release ? [04:18] the top invocation is from the pacakge [04:18] the bottom is the same source [04:23] 2013/04/17 02:50:17 INFO environs/openstack: starting machine 207 in "goscale2" running tools version "1.9.14-precise-amd64" from "https://region-a.geo-1.objects.hpcloudsvc.com/v1/17031369947864/goscale2-1/tools/juju-1.9.14-precise-amd64.tgz" [04:23] 2013/04/17 02:50:17 ERROR worker/provisioner: cannot start instance for machine "207": cannot find image satisfying constraints: failed to get list of flavours [04:23] caused by: https://az-2.region-a.geo-1.compute.hpcloudsvc.com/v1.1/17031369947864/flavors%!(EXTRA string=failed executing the request) [04:23] caused by: Get https://az-2.region-a.geo-1.compute.hpcloudsvc.com/v1.1/17031369947864/flavors: lookup az-2.region-a.geo-1.compute.hpcloudsvc.com: Temporary failure in name resolutio [04:23] after 200 machines, we loose the ability to resolve the name of the hp endpoint [04:24] davecheney, hmm, I seem to be bootstrapping ok in us0east-1 with 1.9.14 from the package [04:24] fwereade: so is it like 6:30am for you now? [04:25] thumper, yeah [04:25] fwereade: cool, not even mind-numbingly early [04:25] fwereade: don't suppose you have time for a quick hangout? [04:25] as predicted I fell asleep early and woke up early [04:25] thumper, sure [04:25] fwereade: i cannot bootstrap into ap-southeast-1 or 2 with that release [04:25] fwereade: I'll start [04:25] but it works perfectly using the source [04:31] * davecheney has had a gutfull of unreliable cloud providers [04:31] ubuntu@juju-hpgoctrl2-machine-0:~$ juju destroy-environment [04:31] error: failed to delete server with serverId: 1510323 [04:31] caused by: https://az-2.region-a.geo-1.compute.hpcloudsvc.com/v1.1/17031369947864/servers/1510323%!(EXTRA string=failed executing the request) [04:31] caused by: remote error: handshake failure [04:50] davecheney, ap-southeast-2 is... well, appearing to bootstrap anyway [04:50] davecheney, but acting a bit weird, I will let you know how I get on [04:56] davecheney, ok, the good news is that the bootstrap node comes up and works fine, the bad news is that the CLI will not acknowledge its existence [04:57] fwereade: -v ? [04:57] did it bootstrap the wrong tools [04:57] ? [04:57] 2013/04/17 06:55:37 INFO environs/ec2: waiting for DNS name(s) of state server instances [i-79c80744] [04:57] 2013/04/17 06:55:42 ERROR command failed: no instances found [04:57] error: no instances found [04:58] eep [04:58] davecheney, the instance definitely exists and it's definitely got the right tools [04:58] fwereade: i;m getting these errors constantly during the load tests [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/2" [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/2" [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/0" [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/0" [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/2" [04:58] davecheney, so that is good anyway [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/2" [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#0#requirer#hadoop-slave/0" [04:58] 2013/04/17 04:57:50 ERROR settings for entity with unrecognized key "r#1#requirer#hadoop-slave/0" [04:59] davecheney, ah ffs there is no way those should be logged [05:00] davecheney, they're not errors, it's just the allwatcher getting its knickers in a twist because it's watching things it doesn't know about [05:01] davecheney, btw, we seem to be waiting only a very few seconds for the DNS name (apse2 issues), which is surprising to me [05:10] fwereade: oh, so that is the api server [05:10] davecheney, yeah [05:10] i was wondering why provisioning didn't land in a screaming heap [05:10] fwereade: i'll log a bug [05:11] davecheney, they're not even really errors -- it's just logging and stopping processing for those docs [05:11] davecheney, cheers [05:11] i'll log two bugs [05:11] can I get a review on https://codereview.appspot.com/8818043/ [05:14] https://bugs.launchpad.net/juju-core/+bug/1169825 [05:21] davecheney, what happened to errors.Newf (if anything)? [05:22] nothign, just people were not calling it correctly [05:22] URL was not a format specifier [05:22] it was arg 1 [05:23] * fwereade sees now [05:23] davecheney, LGTM [05:24] http://paste.ubuntu.com/5715085/ [05:32] https://bugs.launchpad.net/juju-core/+bug/1169826 [06:05] morning [06:07] fwereade: seen my part 1? during sleep I found a by far simpler approach. will continue after breakfast. [06:21] afternoon wallyworld, how's it going today? [06:21] TheMue, great news, I'm afraid I fell asleep last night ;p [06:21] jam1: going well thanks, just sorting out a test isolation issue and i will send a mp for my work === jam1 is now known as jam [06:22] fwereade: no wonder after your night before [06:28] fwereade: second LGTM on https://codereview.appspot.com/8818043/ [06:29] sorry, I meant davecheney ^^ [06:31] bigjools, jtv: Is the maas provider ready to be exposed in jujud/juju client? [06:32] * jtv is not here [06:33] jam: what is the effect of that? Sorry, I have no context on it. [06:34] bigjools: if you have it enabled, people can configure it in ~/.juju/environments.yaml and try to bootstrap against a maas service. [06:34] as far as I know it all works, so if there's something missing I don't know why it's needed [06:34] If it is ready for people to do so, then we should expose it. [06:34] (which it has been) [06:34] But we also want to configure jujud to require it [06:34] jam: we can already bootstrap with it, I don;t understand [06:34] so we don't accidentally lose it later. [06:35] bigjools: I was just confirming that the code has gotten to the "working" state. I hadn't heard a specific update to where you guys were. [06:35] great. [06:35] jam: bzr: ERROR: Cannot lock LockDir(chroot-72851216:///%2Bbranch/goose/.bzr/branch/lock): Transport operation not possible: readonly transport [06:35] I cannot submit that mp [06:35] davecheney: so I would say LGTM trivial on exposing it. [06:35] jam: ok :) yes it should all work [06:35] any ideas ? [06:35] davecheney: you can't use 'lbox submit' for goose. You mark the MP as approved (make sure it has a commit message), and then goose bot will run the tests and land it for you. [06:35] I can do it if you need help. [06:36] ta [06:36] davecheney: how's it going? [06:36] davecheney: so on here: https://code.launchpad.net/~dave-cheney/goose/001-fmt-error/+merge/159287 [06:36] m_3: hey [06:36] there is a "Set commit message" which you can just copy the description if you are happy with it. [06:36] and then mark it Approved. [06:37] sent you an email with the sitrep [06:37] basicallyu can't get above 228 machines [06:37] i susecpt we're running out of ipv4 allocations [06:37] ouch [06:37] and it is being reported in a strange way [06:37] actually, that's awesome... this is the shit we're doing this to flush out [06:38] the way the dns address of the keystone endpoint stops resolving is both [06:38] a. repeatable [06:38] b. fucked up [06:38] * m_3 reading mail [06:39] davecheney: we can see our ip quotas... one sec [06:40] 228 is close enough to 254 that it makes me go hmmm [06:43] m_3: ahh [06:43] interesting [06:43] 228 isn't near any of those numbers [06:44] jam: now what happens ? [06:45] davecheney: hmmmm.... thought that showed ip allocations too [06:45] * m_3 rtfm-ing [06:49] jam: is the bot broken ? [06:50] davecheney: I'll dig through the acct setup with antonio tomorrow [06:51] davecheney: jorge and I are talking tomorrow, so won't get to this til later in my daytime [06:51] davecheney: usually it takes about 5 min for it to wake up and run the test suite [06:51] but if it fails it *should* report a failure to Launchpad and unmark the proposal as pneding [06:52] * davecheney thinks its dead [06:52] davecheney: thanks for the help... /me beddy-bye [06:52] m_3: no worries mate [06:53] davecheney: from what I can tell 'canonistack' thinks the machine is running, but I'm getting 0 response from port 22 [06:56] jam: i remember a discussion a while ago where you metnoined that you weren't able to contact th emachine [06:56] but it appeared to be doing stuff, so whateva [07:14] jam: i'm off to soccer, but here's a wip mp. dimiter was interested to see it. i can't land till his work goes in since i need to account for whatever changes he has made. i also need to land a small goose change [07:14] https://code.launchpad.net/~wallyworld/juju-core/openstack-image-lookup/+merge/159301 [07:15] the covering letter also need to be expanded, but i am late [07:30] * TheMue is happy, the simpler approach works fine. now part two adding the subordinate unite information. [07:54] mornin' all [07:56] rogpeppe2: morning === TheRealMue is now known as TheMue [09:25] in case my earlier email didn't get through, my phone line is down currently [09:25] i'm connected through my mobile currently, but i don't know how reliable it will be [09:33] trivial branch that fixes #1169825 ; review appreciated please: https://codereview.appspot.com/8821043 [09:58] YEEEEEAH! [09:59] * TheMue does the subordinate dance ... [10:09] fwereade: i'm after a review of this fairly trivial branch, please. i think it's important to get in before the release. https://codereview.appspot.com/8823043 [10:09] rogpeppe2, ok, I'll chat about s3 in a mo [10:10] fwereade: ok [10:11] rogpeppe2, nice, LGTM [10:12] fwereade: trivial? [10:12] fwereade: you've got a proposal :) [10:12] rogpeppe2, yeah, I think so [10:12] fwereade: cool, thanks [10:12] TheMue, cheers [10:12] * rogpeppe2 wonders how long it'll take to run lbox submit through the mobile data connection [10:13] fwereade: it's https://codereview.appspot.com/8824043/ === rogpeppe2 is now known as rogpeppe [10:13] TheMue, cheers [10:13] rogpeppe, I am somewhat ambivalent about changing Storage.Put [10:14] fwereade: i think that's going to be a necessary thing if we ever want retries to work. [10:15] rogpeppe, I would be much happier if it took a Reader and tried to ReadSeeker it [10:15] fwereade: i wondered about that, but i was very glad i didn't [10:15] rogpeppe, oh yes? [10:16] fwereade: because the static type caught lots of places that i would not otherwise have found [10:17] fwereade: it's not that we can stream data without having it first either [10:17] fwereade: because we need to know the length [10:17] rogpeppe, it is not unheard of for content lengths to be declared [10:17] fwereade: that is true, i suppose [10:18] rogpeppe, if they are, then great,but even if not I don't really think we should force all clients to buffer everything [10:18] fwereade: but in all the cases we care about, we do want retries to work [10:18] fwereade: and they won't work on a straight io.Reader [10:19] rogpeppe, maybe it's our own responsibility to buffer though [10:19] fwereade: i think so [10:19] rogpeppe, ha, different "our" I suspect [10:19] fwereade: ah, maybe so :-) [10:19] fwereade: i really don't think that Put should do the buffering itself [10:20] fwereade: if that's what you were thinking of [10:20] rogpeppe, yeah, I definitely see that side of it too [10:20] rogpeppe, but I'm not sure it's a very strong objection -- convince me? [10:21] fwereade: in many of the places, we already have the data in a file [10:21] rogpeppe, in which case, yay, easy ReadSeeker [10:21]