BradBmaybe we'll just new a few maintainers (and more of you, of course) to test it offline and then just take the leap at some point in the next few weeks12:01
BradBs/new a/need a/12:01
mdzanother possibility would be to have pitti use it for tracking security issues only12:02
mdzsecurity is the toughest test case for a bug tracker :-)12:02
mdzthat has the advantage of being a relatively discrete set of bugs12:03
mdzworkrave, back in 10 if you still need me12:04
BradBok, thanks for your feedback12:04
debonziBradB, is it possible to have a debug port in launchpad.ubuntu.com?12:06
BradBelmo: ^ ? (debug is on 9021)12:08
kiko_bzis it now?12:08
BradBkiko_bz: yeah12:09
kiko_bzah, neat.12:09
BradBit won't work though unless elmo does something to the apache config12:09
debonziBradB, sure? I can't connect in port 902112:09
BradBdebonzi: yep it's listening. the firewall'll block you, of course, but it's listening :)12:10
kiko_bzscreaming on deaf ears12:10
debonziis elmo the responsable for that?12:10
BradBdebonzi: yep12:11
debonzielmo, ?12:11
kiko_bzBradB, cprov has an updated gina run for your schema version I believe12:13
kiko_bzI don't know how you would go about using it though, given you've already got some data in your DB12:14
BradBkiko_bz: sounds like a problem for the stud12:14
BradBin any case12:16
kiko_bzone thing we *could* do is start a run on *your* database if you like.12:16
kiko_bzit will take 2h.12:16
BradBi gotta go play the role of gigalo now12:16
kiko_bz(just for main)12:16
kiko_bzhe ran it on soyuz_dogfood12:16
BradBkiko_bz: will running it on launchpad_dogfood blow any data way? i wouldn't expect so. if not, go ahead, i'd say.12:17
BradBi gotta go out for a while.12:18
=== BradB is now known as BradB|out
kiko_bzit probably won't, but I don't know if it will run :)12:18
cprovBradB|out: heeh it's your last chance to say "please NO" ...12:18
kiko_bzcprov, what do you think :)12:19
kiko_bzcprov, make a dump of the current launchpad_dogfood DB in case it goes wrong.12:19
cprovFOGO NA BOMBA !!12:19
kiko_bzyou can actually dump your DB with inserts only12:19
kiko_bzwhich would avoid having to re-run gina12:19
kiko_bzI don't know if that would blow up because of uniqueness constraints though12:19
kiko_bzmay be simpler to just kick gina again12:19
BradB|outif you think you might blow away data, just make sure you have a quick way of unblowing it away12:19
BradB|outif you need more input from me though, it'll have to wait, i gotta head out nowish12:20
cprovjust dumping the current launcpad_dogfood should avoid future problems12:20
cprovthe only problem is Librarian  dependencies ( Kinnison WANTED), once I'm running w/ librarian there is no way to insert its fields latter, so we will need to purge the DB and rerun gina+librarian12:22
sabdflok, am back, will catch up on scrollback12:22
=== salgado [~salgado@200-206-134-238.async.com.br] has joined #launchpad
kiko_bzsabdfl, meet salgado; he's working with us initially on setting up some agreeable sampledata for soyuz12:24
sabdflhey salgado, welcome aboard12:25
salgadohello sabdfl12:25
kiko_bzthe current size we've paired it down to is 1.4MB bzipped12:25
kiko_bzI suspect that's still too big12:25
sabdflhow many packages would that include?12:25
sabdflwe don't need many12:26
sabdfljust four or so packages with four releases of each12:26
salgado121 sourcepackages12:26
kiko_bzhow many releases?12:26
salgado303 binarypackages12:27
kiko_bzwhere is all that size coming from, damn it? people?12:27
salgado50 people12:27
kiko_bzit's a reasonable sample actually12:27
kiko_bzsalgado, from a visual review of the dumpfile, what's taking up most of the space?12:27
salgadoI haven't take a look at it yet. I'll do it now12:28
salgado660K from bziped gpgkeys12:30
kiko_bzsalgado, can you kick all but 50 of those gpgkeys spacewards? I don't care for them, and I suspect sabdfl doesn't either.12:31
salgadoproject/product descriptions should be blamed too12:31
kiko_bzblame cprov, he got the data for us ;)12:32
salgadokiko_bz: there's 20 gpgkeys12:33
cprovsalgado: have you sliced the packages.gz files ?12:34
=== kiko_bz is now known as kiko
cprovsalgado: or are following another approach ?12:35
salgadocprov: I used cascade=True on all ForeignKeys() and then I deleted all but the first 50 Persons12:35
cprovsalgado: so, you worked just on DB not in the code itself, fine12:36
cprovkiko: next step ? store the new dump in a public place ? or turn it our new-sampledata for lp ? 12:42
kikoI would really like to put this up as the new sample data12:42
kikobut it needs to be a bit smaller12:42
kikoI think ~800K uncompressed is the limit of what's acceptable12:42
sabdflstub: up early? are bug attachments up and running?12:58
cprovkiko: then cut it a bit more, let's say 20 persons12:58
kikosalgado, how does that sound?12:58
salgadofor 800k uncompressed I think 10 persons is enough12:58
kikoshoot then 12:58
=== doko [doko@dsl-084-057-020-211.arcor-ip.net] has joined #launchpad
dilysMerge to rocketfuel@canonical.com/launchpad--devel--0: template fix to avoid traceback when SourcePackage.distro is NULL. (patch-707)01:12
kikodebonzi rocks01:16
dilysMerge to thelove@canonical.com/dists--tla--1.3: seal the branch, not used now the name has been chosen (version-0)01:23
sabdfli think malone is a lot slicker than most bug trackers already, from an end-user point of view01:24
sabdfli want an issue tracker so that as many of the "doh" things as possible are handled by non-developer community members01:24
sabdflbut i'm not going to hide malone01:24
sabdfljust led users to the issue tracker first01:24
dilysMerge to thelove@canonical.com/tla-debian--debian--1.0: seal the branch, not used now the name has been chosen (version-0)01:24
dilysMerge to thelove@canonical.com/tla--devo--1.3: seal the branch, not used now the name has been chosen (version-0)01:25
dilysMerge to thelove@canonical.com/docs-tla--devo--1.3: seal the branch, not used now the name has been chosen (version-0)01:25
sabdflthe ui weirdness was debated between limi and i01:25
sabdflwe decided to try it, we can pull it if it's too weird01:26
sabdflthis is the bit where the headline assignment is a rownavigation style table01:26
sabdflwith links in it, on the package/product and assignee01:26
sabdfland the links take you to different pages than clicking elsewhere on the row01:26
sabdflit's a handy shortcut if you know what you're doing01:26
sabdflbut might be too clever for new users01:26
salgadokiko: 10 persons, 36 products, 13 sourcepackagereleases: 900K uncompressed01:44
kikosalgado, looks good; let me check it out01:45
kikosabdfl, is 900K closer to acceptable? it triplicates the current data.01:45
sabdflfogo na bomba indeed :-)01:48
kikolet's see01:48
sabdflkiko! you got your name back!02:00
sabdflnight guys02:09
salgadonight sabdfl02:09
sabdflhey sync-man!02:09
sabdfllifeless: we vending our branches yet?02:10
lifelessnot quite. taxi had a quirk beyond what spiv fixed.02:10
lifelesson my plate for today - I've done the bootstrap heavy lifting for baz now.02:10
lifelessyou get roll a deb of baz to play with anytime now :)02:11
sabdfli saw thelove commits sealing off tla02:11
sabdflgotta love thelove02:11
sabdfli crack myself up02:11
sabdflaren't you guys glad we don't work in the same office every day :-)02:12
=== stub yawns
dilysMerge to rocketfuel@canonical.com/arch-pqm--main--0: dont fail pqm if a PQMException is raised (patch-12)03:44
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: * added directories (patch-3)03:55
=== doko [doko@dsl-084-057-106-014.arcor-ip.net] has joined #launchpad
=== kiko_ [~kiko@200-206-134-238.async.com.br] has joined #launchpad
=== kiko_ [~kiko@200-206-134-238.async.com.br] has joined #launchpad
=== BradB|out is now known as BradB|zzz
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Re-adding the massively pruned patch logs (patch-4)07:20
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Nudge the tla help command to the left (Jivera) (patch-5)09:02
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Unsigned checking for Asuffield (Jivera) (patch-6)09:51
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: I believe the update test is finally working (patch-7)09:55
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Adding grab file description to help (patch-8)10:01
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Fixing grab so that it registeres the remote archive properly (patch-9)10:05
=== lulu [~lu@host217-37-231-28.in-addr.btopenworld.com] has joined #launchpad
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Various small things, mostly input validations (Abentley) (patch-10)10:20
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Selective undo (Monoses) (patch-11)10:40
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Path fix for cmd-file-find (patch-12)10:45
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Add missing patchlogs from before (patch-13)11:01
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Fix for get-changeset and testcase for same (Bentley) (patch-14)11:06
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Make tests more solaris friendly (patch-15)11:13
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: Fixing id for test-get-changeset (patch-16)11:19
=== ddaa [~ddaa@nemesis.xlii.org] has joined #launchpad
sabdfllifeless: is there a generic command to show the diffs of the last commit?11:50
spiv_Maybe something like "tla get-changeset $(tla logs -r | head -1) ,,last-changeset && find ,,last-changeset/patches -type f | xargs cat"11:52
sabdflstub: what do the new reltriggers do?11:52
spiv_There may be a saner way to do that :)11:53
sabdflspiv_: yowser11:53
stubNew reltriggers?11:53
sabdflstub: from this morning's refuel:11:53
sabdfl@@ -676,6 +679,7 @@11:54
sabdfl UPDATE pg_catalog.pg_class SET reltriggers = 0 WHERE oid = 'sourcepackagerelease'::pg_catalog.regclass;11:54
sabdflto current.sql11:54
sabdflah, not new11:54
stubI think that is the code pg_dump puts in there to turn of relational integrity constraints and triggers11:55
stubWhich needs to be done now, as our sample data has grown in complexity11:56
spiv_sabdfl: This is probably a little saner: "tla delta --diffs $(tla logs | tail -2)"11:57
stubsabdfl: You should be able to use archzoom if someone fixes the gpg signing rules for the account it runs as.12:00
sabdflspiv_: that did the trick nicely, thanks12:02
=== carlos [~carlos@69.Red-80-33-181.pooles.rima-tde.net] has joined #launchpad
sabdflhey carlos12:05
carlossabdfl: hi12:05
carlossabdfl: could you look at my mail about the statistics and fuzzy flag from yesterday at launchpad mailing list?12:39
sabdflcarlos: ok12:39
carlossabdfl: thanks12:39
=== Kinnison attempts to make a short list of everything gina needs to do. I've been reading through the source to familiarise myself with it.
Kinnisonsabdfl: https://chinstrap.warthogs.hbd.com/~dsilvers/gina.txt01:32
=== Kinnison thinks that is all we need for dogfooding; the last one probably isn't needed
sabdflKinnison: only the first one is critical01:33
sabdflsingle suite, sinlge arch would be fine01:33
sabdfluse hoary01:33
sabdflit's getting uploads01:33
Kinnisonsingle suite single arch still needs the overrides and removals stuff01:33
Kinnisonotherwise our copy of hoary won't get cleaned up01:33
sabdflcan't just drop the publishing table each run?01:34
KinnisonI guess I could do it that way if you'd prefer01:34
KinnisonThat'll get us up and running faster I guess01:34
KinnisonI'll do that01:34
=== Kinnison fiddles a bit
Kinnisonhow about now?01:35
KinnisonOkay; that's what I'm aiming for with gina01:36
=== Kinnison is going to go and find some lunch (I forfeited breakfast in favour of reading code)
KinnisonThen I'll start01:37
KinnisonCould I ask you; as a side-line; to look at the database/publishing.py and related files and suggest a split which would be sane? I've not got the hang of predicting these things yet :-)01:37
=== Kinnison -> lunch
ddaatrying to use str.translate on a unicode string gives an error.01:45
ddaaCan somebody point me at the right thing to use? That may save me some doc-searching time.01:46
=== lulu [~lu@host217-37-231-28.in-addr.btopenworld.com] has joined #launchpad
=== ddaa finds a hack appropriate for the situation
stubddaa: use unicode.translate instead?01:57
ddaaHa... not quite I was doing...01:57
ddaaInstead I was simply encoding the input to utf8.01:57
ddaacontext: I'm writing a job name munging tool to work around the broken shell quoting in cscvs.01:58
ddaaThe error was: exceptions.TypeError: character mapping must return integer, None or unicode01:59
ddaaHo... apparently it's already using unicode.translate since the output is unicode... the trick seems to be it accepts a unicode string as translation table and I gave a plain string...02:02
ddaastub: thanks02:02
ddaaduh... now that I handle the unicode string okay, it's twistd which complains because it gets a unicode instead of a str...02:07
Kinnisonunicodeobject.encode('utf-8') -> str object02:07
Kinnisonstr.decode('utf-8') -> unicode object02:07
stubbad twisted!02:07
ddaaKinnison: ha, simpler than "import codecs; codecs.getencoder('utf8')(text)"02:08
Kinnisonddaa: aye02:08
=== Kinnison uses thingstr = footable.barcolumn.encode('utf-8') all over the place in his db code
ddaaAfter all, I'm going utf8ize it, then...02:08
ddaaOkay... that works.02:11
ddaaI mean, I'm hitting another bug, but that part works...02:12
=== cprov [~cprov@] has joined #launchpad
Kinnisonsabdfl: Section and Component don't have owner columns -- are they officially shared among all the distributions then?02:40
sabdflKinnison: yes02:41
Kinnisonsabdfl: Cool; any complaints if I propose making them unique then?02:41
sabdflah. very good thinking02:41
Kinnisonhmm; someone appears to have beaten me to it02:42
KinnisonI really must get gina importing into the new db so I stop asking these things :-)02:42
=== Kinnison ponders some more sleep in the meantime :-)
cprovKinnison: hi, can we get Librarian running on mawson ?02:45
Kinnisoncprov: I started one yesterday I'm sure02:45
=== Kinnison looks
ddaalifeless: you are probably asleep, but I straced a hang at "launched local CVS server"02:45
Kinnisondsilvers@rosetta:~ $ ps ax | grep libra02:46
Kinnison 6976 ?        S      0:00 /bin/sh ./run-librarian.sh02:46
Kinnison 6977 ?        S      0:00 /usr/bin/python2.3 /usr/bin/twistd -noy /srv/launchpad.ubuntu.com/launchpad-dogfood/lib/canonical/librarian/server.tac02:46
ddaalifeless: one of the threads hangs on select, the other hangs on futex.02:46
lifelessddaa: mail it to me ?02:46
ddaaBtw, I'm in process of trying to test taxi.py...02:46
lifelessthank you02:46
lifelessI was going to do that today, ended up bazzing all day02:47
ddaaBut roadblocks keep popping up...02:47
ddaaI have a "munge job names" hack ready. I want to test it over the whole process before merging.02:48
ddaalifeless: the robertcollins.net mail is still okay?02:48
cprovKinnison: I had already ran gina successfully on mawson, but it is unusefull w/o Librarian, I think02:48
lifelessddaa: yes02:48
Kinnisoncprov: librarian is running02:48
Kinnisoncprov: the uploadfactory is on port 909002:49
Kinnisoncprov: if I star-merge will I get your gina changes to make her run on mawson?02:49
cprovKinnison: fine, btw, I'm working on CURRENT srcpkgrelease replacement02:50
cprovKinnison: just few minutes, I will request merge for PQM02:50
Kinnisoncprov: cool; thanks02:51
ddaalifeless: debugging data sent02:54
ddaaI'll stand by if you want additional in-vivo data.02:54
lifelessddaa: debug-bomb the native protocol please.02:54
ddaalifeless: you really should bump down the compression...02:55
ddaa"debug-bomb"? "native protocol"?02:55
lifelessstart with hasProtocol() and follow your nose02:56
lifelessthere is a cvs client implementation02:56
ddaaokay, got the native part...02:56
lifelessif its hung there, then its reading from an empty pipe.02:56
ddaayou want me to spill printlines all over the place, right?02:56
lifelesskilling the cvs server should resume the process.02:56
ddaaThere is no cvs process.02:56
ddaa"pgrep cvs" gives nothing.02:57
lifelessself.logger().debug statements go to the log, and are useful.02:57
lifelessno cvs process - then what is the select selecting on ?02:57
ddaayou tell me...02:57
lifelesswhat fd ?02:57
ddaaselect(6, [0 5] , [] , [] , NULL <unfinished ...>02:58
lifelessin the strace, what was opened on that fd most recently ?02:58
=== ddaa checks if strace has a --time-travel option
ddaasorry boss, the strace folks do not yet support tracing system calls before the process is attached.03:04
ddaaI suspect this feature is not on top of their todo...03:05
ddaaspeculation: spawning cvs has failed for some reason only known to cvs. The twisted stuff has a intercepted the sighup. The thread is blocking trying to read on a pipe for a dead process.03:15
ddaaor something equally crackful... 03:16
lifelessddaa: no, the spawn succeeded03:17
lifeless 2004/10/28 14:15 CEST [Broker,client]  WARNING:root:launched03:17
lifeless        local CVS server03:17
lifeless        for :local:/home/david/canonical/launchpad/botmaster/slave/buildbot-jobs/vte_HEAD_import.job/vte@arch.ubuntu.com/vte--MAIN--0/cvs_temp_repo03:17
lifelessthat is logged stricly after the the pipes have been created.03:17
ddaasure, that does not mean cvs did not die immediately after.03:17
lifelessit might be a lost signal03:18
ddaaThat would be consistent with the insanity that led to arch._twisted.03:18
lifelessso what you need to do is to run strace from before you start the job.03:18
ddaaimho that not really working around a pyarch bug, just working are twisted's lack of tolerance.03:18
lifelessddaa: iynsho.03:19
=== ddaa dicts
lifelessits actually a unix process model problem.03:19
lifelessIn Your Not So Humble Opinion03:19
dilysMerge to rocketfuel@canonical.com/launchpad--devel--0: missed clauseTables, people traversed by name, gina/nicole running on mawson (patch-708)03:19
ddaaWell, I may be wrong.03:19
ddaaThat's speculation.03:20
lifelessactually, it can't be a dropped signal03:20
lifelessif the signal is recieved, the pipe is closed at the far end, read will generate SIGPIPE03:20
lifelessteisted doesn't AFAIK install SIGPIPE handlers03:20
lifelessddaa: can you please try this:03:21
ddaawhatever it does, the effect I observer are undistinguishable from that.03:21
lifelesstla register-archive03:21
dilysNew bug 2150 for Launchpad/Rosetta: Fix the way we store fuzzy strings in the database schema03:21
ddaalifeless: whose key is used there?03:22
lifelessgood point.03:22
cprovKinnison: new gina merged on RF, please try .03:24
ddaalifeless: you want be to build and install baz, right?03:24
lifelessjust to check that that archive location is abrowse -s 'able for you03:24
lifelessis up and going now03:25
lifelesscause thom is a legend03:25
lifelessnay, a god even.03:25
Kinnisoncprov: merged already?03:25
ddaaLooks like abrowse -s is happy.03:25
cprovKinnison: yep03:26
=== ddaa looks on the Bazaar page for an apt-sources line
Kinnisoncprov: dilys hasn't mentioned it; hence my confustion03:27
cprovKinnison: let's plan something for today, we have to run gina and nicole on mawson, definitly, as milestone :)03:27
Kinnisoncprov: I concur03:28
cprovKinnison: dilys doesn't like me :)03:28
=== Kinnison is doing some gina changes now that sabdfl and I agreed upon
=== Kinnison workraves; brb
ddaalifeless: any idea what the apt-sources line should look like?03:29
sabdfllifeless: how are we looking for the automatic sync-testing on galapagos?03:29
cprovKinnison: btw, do you have something ready for publishing the query for and update the last CURRENT srcpkgrelease to SUPERCCED before insert a new one, I mean, true publishing system for gina ...03:30
lifelessddaa: its not a repository03:31
lifelessbut I believe you can just do deb  path  ./03:32
ddaaNo good. I want crack-of-the-day, not crack-of-the-last-time-I-visited-the-page.03:32
ddaa"path" being what?03:33
lifelesssabdfl: haven't heard from james re the account for that. code wise, I want to get the production drop onto macquarie and galapagos done tomorrow, which updates tazi03:33
lifelessand then I'll start on the unattended-scheduling, which I'm estimating @ a couple of days03:33
sabdflvendor, default branches, done? you're running out of space between 95% and 100% :-)03:34
lifelessno progress at all since last phone call.03:34
Kinnisoncprov: sabdfl and I agreed that initially gina will just totally repopulate the publishing tables each run03:34
=== kiko [~kiko@200-206-134-238.async.com.br] has joined #launchpad
lifelessbeen baz bootstrapping03:34
Kinnisoncprov: over time; I'll write the code to do superceded etc and at that point we can add support for it into gina03:35
Kinnisoncprov: for now; sabdfl and I agreed that we just wanted daily imports working :-)03:35
Kinnisoncprov: see https://chinstrap.warthogs.hbd.com/~dsilvers/gina.txt for the plan :-)03:35
cprovKinnison: fine03:35
ddaaAnybody aware of a debug-enabled variant of popen2.Popen3?03:36
Kinnisoncprov: what did you change in grabber.py ?03:37
cprovKinnison: nothing .. I just change back some particular conf 03:38
sabdflcprov, Kinnison, it's ok for the moment if it just shows the current set as published, and all older ones are just "there"03:38
cprovKinnison: which pass to use on https://chinstrap ?03:39
sabdflin the sense that you can see old releases of a source package in the system, but not superceded info03:39
sabdflfor the moment03:39
Kinnisonsabdfl: yeah; that'd be the side-effect if we blat the publishing records each time03:39
sabdflno problem for this week03:39
Kinnisoncprov: erm, wartyhoa... I think03:39
cprovKinnison: sabdfl: I'm just affaid to have two or more CURRENT srcpkgrelease03:40
lifelessddaa: anyway, as chinstrap isn't ppc, you need to roll your own anyway :)03:40
Kinnisoncprov: define 'current' if you mean in the publishing tables then that won't happen03:40
Kinnisoncprov: not until we do things properly :-)03:40
ddaalifeless: what does ppc has to do with that???03:40
lifelessoh right, you have an intel desktop, my bad.03:40
cprovKinnison: I do mean publishing tables, yes, gina always insert new packages as published, if I have a new release of it, it will be inserted as published too03:42
Kinnisoncprov: yeah; but we're going to empty the publishing tables before the gina run03:42
Kinnisoncurrent gina gives me:03:42
Kinnisonlibpq.OperationalError: ERROR:  insert or update on table "sourcepackagerelease" violates foreign key constraint "$4"03:42
KinnisonDETAIL:  Key (component)=(3) is not present in table "label".03:42
Kinnisonagainst the current schema03:42
Kinnisonsomehow the sourcepackagerelease component column has been pointed at the label table03:43
=== Kinnison boggles
Kinnisonstub: ping?03:45
cprovKinnison: are you running over a DB created by lp-sampledata ?03:45
cprovKinnison: it is looking for 'restricted' component03:45
Kinnisoncprov: No; just a plain 'make' in the schema dir for now03:45
stubKinnison: That dates back to the dawn of the database (database/schema/archives/launchpad-1-00-0.sql)03:46
=== Kinnison wonders how this ever worked
Kinnisoncprov: actually; universe :-)03:46
cprovKinnison: that is it ...03:46
Kinnisonstub: can you please commit a fix to make that point at component(id) instead of label(id) ?03:47
cprovKinnison: do you have the component entry ? wierd03:47
stubIt is nullable - I guess people just stick null in it. I think it must have been from an old use of the label table. I assume it is supposed to reference the component table?03:47
Kinnisonstub: yes please03:48
=== Kinnison will fix it in his testdb for now; but if you could commit a real fix I'd appreciate it
Kinnisonthere must have just been entries in the label table before now03:49
ddaawhat's the extra bit needed to enable the output of self.logger().debug?03:49
ddaalifeless: ^03:49
Kinnisonyay: libpq.OperationalError: ERROR:  current transaction is aborted, commands ignored until end of transaction block03:51
lifeless#       TODO convenience for interactive debugging move somewhere importable.03:51
lifeless#        import logging03:51
lifeless#        logging.basicConfig()03:51
lifeless#        logging.getLogger().setLevel(logging.CRITICAL)03:51
lifeless#        logger=logging.getLogger('a')03:51
lifeless#        logger.setLevel(logging.DEBUG)03:51
lifeless#        logger.debug('fooooo')03:51
lifelessthats in test_CMDS.py03:51
lifelesswhen I needed to do something similar03:51
lifelessin buildbot, if you are running the slave as normal, you can just edit buildbot_slavecommand.py03:52
lifelessand tweak the level there.03:52
=== ddaa wonders if the buildbot types ever heard of "configuration"...
lifelessddaa: there is a todo to pass this down as part of the job.03:53
lifelessits not intended to be per-slave.03:53
ddaaMh... the self.debug=0 bit i guess...03:56
lifelessplease, just change your slave level.03:57
=== debonzi [~debonzi@] has joined #launchpad
ddaawhat ... is ... a ... slave ... level ...03:58
=== BradB|zzz is now known as BradB
lifelessin buildbot_slavecommand.py03:58
lifelesslook for WARNING03:58
ddaaha... was looking at the wrong file...03:59
ddaawas looking at buildbot/slavecommand.py...03:59
lifelessheh. we want importd/buildbot_slavecommand.py :)03:59
=== BradB puts coffee on
ddaaokay, I'll keep that around when building. That will probably prove useful.04:04
ddaaOf course, this time the server is not...04:04
ddaa(me stops and starts the import from scratch, just for kicks)04:04
KinnisonOkay; zhongshan has something broken with its cursors or something. disabling them helps a lit04:07
Kinnisoncprov: I have a gina running now against the new schema; with my build record fixes04:07
Kinnisoncprov: I'll let you know how it goes04:07
cprovKinnison: you rock !!04:08
Kinnisoncprov: currently running on zhongshan because I have good debug-ability there04:08
Kinnisoncprov: I'll run it on mawson once I'm satisfied :-)04:08
cprovKinnison: ehe zhongshan is gives a pretty better environment the mawson :)04:11
cprovKinnison: starting with this crap colored prompt <wink>04:12
kikoman I am SLEEPY today04:13
Kinnisonkiko: I know what you mean04:13
=== Kinnison ponders the elixir of life
lifelessyay, it failed to build.04:21
Kinnisoncprov: I think it's about 10% of the way through the import04:26
cprovKinnison: everything working ?? I mean Librarian and so on ... are you importing just main and restricted ?04:27
Kinnisoncprov: everything working; I'm importing it all for a test04:29
Kinnisoncprov: I'm about to go prodding at the build table (I have a perioding commit in gina so if she stops; she doesn't lose all her hard work)04:29
=== Kinnison worries
Kinnisondsilvers@petitemort:~ $ ssh snogshan04:30
Kinnisonssh: snogshan: Name or service not known04:30
Kinnisonwhere did *that* misspelling come from?04:30
Kinnisonboo hiss; my build patch didn't work04:32
=== Kinnison kills gina
=== debonzi [~debonzi@] has joined #launchpad
ddaaCanonical Productions presents, a new entertainment series on your channel: "Who Killed Gina". Starring: Ubuntu, Linux, Zope, Twisted, Plone...04:39
Kinnison...and Daniel (the wider)04:39
ddaa... The Sushi Place Near Mark's...04:39
lifelessand cartman04:39
carloscprov: ok, I have the .sql you gave me imported into the database, do you have it already connected with our arch repository?04:39
=== ddaa raises on eyebrow
=== ddaa googles cartman
carlosI mean, from the package information, how could I construct a "tla get" command?04:40
cprovcarlos: not yet, but Kinnison is working on it 04:40
carloshmm, is there a bug report for that task?04:41
ddaalifeless: ha... I meant something else, but nm :-)04:41
ddaaSo... pyarch/baz04:42
carlosKinnison: ?04:42
Kinnisonwhomewhat where?04:43
=== Kinnison hunts for his teflon shoulderpads
carlosKinnison: about the soyuz <-> arch connection04:43
ddaaPython way to look for a command in the path w/o running it? Get the exit status of sh -c 'which baz > /dev/null', or is there something more pythonic?04:43
carlosKinnison: cprov said you are working on it, is there a bug report for that task so I can be notified when it's ready?04:44
Kinnisoncarlos: why would I be working on that?04:44
carlosKinnison: cprov say that :-P04:44
Kinnisoncarlos: *KEYBUK* is doing arch stuff; dunno who's doing the soyuz integration :-)04:44
Kinnisoncprov: stop giving me new work to do or I shall fly out to brazil and tickle you04:44
carloslifeless: Do you know anything about the soyuz/arch integration?04:47
lifelesscarlos: keybuk04:47
carlosperfect, thanks04:48
=== ddaa ponders how he's going to make that fit with the current spawning strategy stuff...
cprovKinnison: you are welcome to see waht is waiting for you :)04:50
ddaaHow about adding a arch.config module which stores the spawing strategy name, the command name (and maybe other things) and a .config()04:55
ddaaand a .configure() method in various modules. How does that sound?04:55
lifelessddaa: context ?04:56
ddaaWant to make the name of the command spawned for arch operations configurable.04:56
ddaaThe current spawning strategy stuff works with an object which gives no way to modify that after the object is created.04:57
ddaaBecause spawning strategies store the command name. No good reason for that though...04:57
ddaaI cannot be a module variable though because it would make testing problematic.04:58
ddaaHa yeah, the name is stored in the strategy to avoid redundant data. The command name is given only once at object creation.04:59
ddaaAn easy way is running the command-name tests at module import-time, when the strategy is created, but that sucks, because it gives no clean to customize the command name.05:00
ddaaSo, considering my recent canonical.lp experience, I'd rather move the configuration stuff in a separate, low-level module.05:01
=== debonzi is now known as debonzi|lunch
ddaaThere is also the fact that the command name should be configurable easily w/o having to import arch._tla05:01
ddaaTherefore arch.config sounds good.05:01
lifelesshaving the name is fine, IMO, just have multiple instances of the strategy/.05:02
lifelesswhat prevents that ?05:02
ddaaI do not parse you: "having the name is fine"05:02
lifelessin the instance.05:02
lifeless'this is a spawning strategy for tla'05:02
=== cprov -> lunch
ddaaYeah exactly.05:02
lifeless'this is a spawning strategy for cvs'05:02
lifelessso I don't think you need to change anything.05:03
ddaaThat's it's meant to be.05:03
lifelessor are you looking at how to detect baz ?05:03
ddaaActually, I want to decouple command name choice from spawning strategy choice.05:03
lifelessso refactoring this is easy....05:04
lifelessstart be replacing all hardcoded 'tla' that needs to change with a call to a method05:04
lifelesshardcode that method to return 'tla'05:04
Kinnison builds05:04
Kinnison     4405:04
Kinnison packages05:05
Kinnison       7105:05
lifelessthat method can be a module level one in this case, as you don't really have an appropiate class handy.05:05
=== ddaa withhold objections until he sees the end
lifelessso this is correctness preserving.05:05
=== Kinnison has gina seemingly preserving the connection between the builds of binary packages :-)
ddaalifeless: EOF05:05
lifelessstep two. in that method, put your policy for determining the name. (i.e. search the path for baz, then for tla)05:05
lifelessthis adds baz support.05:06
ddaaYeah, that's the obvious thing to do so far...05:06
lifelessbut: knowing the name is baz won't be enough over time05:06
lifelessyou need to couple the tla version information you have with the name.05:06
=== ddaa winces
Kinnisonlifeless: will baz have the cacherev/revlib fix?05:07
lifelessKinnison: eh?05:07
ddaalifeless: you are done?05:07
Kinnisonlifeless: the one to make revlibs use cacherevs; or is that in already?05:07
lifelessddaa: well no, but you see where I'm going05:07
lifelessI think05:07
lifelessKinnison: don't know, right now, 1am, don't care. :)05:07
ddaaNow the pb is that 'tla' is hard coded in only one place. At the top-level of arch._tla and it reads "spawn = DelayedGuessedSpawningStrategy('tla')"05:07
lifelessmost importantly, we will be bugfixing everythng we can lay our hangs on.05:07
Kinnisonlifeless: fair enough05:08
lifelessso a question like 'will baz fix foo' - the answer is 'duh'.05:08
=== Kinnison giggles
ddaawhich means that going down the road you point does not make it possible to decouple spawning strategy and command name.05:08
lifelessddaa: well, I'm suggesting they be completely decoupled. you have a command description object, and a spawn strategy object, and one uses the other.05:09
ddaaExcept if there is a way to rebuild the command strategy with the right name w/o requiring knowledge of either. That would be the job of the arch.configure()05:09
ddaaYeah. I see.05:10
ddaaJust said that way it makes sense.05:10
lifelessddaa: I'm fading, can't focus well enough - but arch.configure feels wrong to me - wrong enough not to ignore.05:10
ddaaSo... it's about mutating the command description object, so the spawning object needs not know anything, right?05:11
ddaaAnd giving a reference at some module level to access the command desc object.05:12
lifelessyou just need a new command description object instance for baz, and select that when you want it.05:12
lifelessi.e. arch.implemention05:12
lifelesswe're saying the same thing05:13
ddaaThat's the "select that" part which is causing me trouble.05:13
ddaaFrom the start.05:13
ddaaThe spawning strategy gives no access to the command desc so far.05:14
lifelessthats what I was stepping you through refactoring05:14
ddaaWhich I could change, but I sorta like that way because it made DelayedGuessedSpawningStrategy transparent...05:14
lifelessit would stay transparent. this is api preserving.05:15
=== ddaa rereads the refactoring steps
lifelesstell you what. this might be best shown collaboratively. I can show you on a shared screen session, or in irc pastes, after I sleep05:16
ddaaGo bed pal. I need to sit on it for a bit.05:16
ddaamh... the arch._arch thing turns out it was a bad idea after all...05:24
ddaashould be something like... arch.implementation...05:24
lifelessoh yeah.05:28
ddaaarch.backend would be a better name though.05:28
dilysMerge to thelove@canonical.com/bazaar--devo--1.0: add unit test dead chickens to libarch (patch-17)05:28
lifelessauto updates on commit :_)05:28
lifelessget your crack of the minute right here folk05:29
=== ddaa thinks the server clock should be set to UTC...
=== lifeless could care less
lifelessget your debs here baby!05:30
ddaa16:28... have to know it's the UK tz.05:30
lifelessdude, it so doesn't matter.05:30
lifelessthe http timestamp is in UTC regardless.05:30
ddaaI'd have to annoy tom to he makes that a proper repository.05:31
lifelesstry this as an apt repo:05:31
=== ddaa runs to /etc/apt/sources.list
lifelessdeb http://bazaar.canonical.com/packages/debs/ ./05:31
=== ddaa sigsevs aptitude
ddaamaybe it's not a repo, after all.05:34
lifelessI'll need to create a Packages file05:34
=== ddaa forgot the "deb " prefix
=== ddaa goes back to brood over arch.backend in the meantime
=== carlos goes to the university
carlossee you later05:38
ddaaproperties are cute... but they blow the seamlessness of modules and instances.05:39
=== ddaa wants module properties
lifelessyet another reason that modules should simply be class instnaces05:42
lifelessor better yet, instances of 'namespace', and do away with the whole module nonsense :)05:42
ddaasounds like smalltalk :-)05:42
Kinnisononly if you get rid of the class concept too and have namespaces only please05:42
spiv_ddaa: Just insert an instance into sys.modules directly.05:42
lifelessKinnison: you lisper at heart05:43
=== ddaa sees the winkle and puts down the lart
=== lifeless shakes at kinnison
Kinnison(state 'pah (lifeless))05:43
lifelesslost me05:44
Kinnisonrandom broken lispish equivalent of 'lifeless: pah'05:44
Kinnisonpah -> lifeless05:45
Kinnisonor however you'd do it in smalltalk05:45
ddaaThat sounds good so far, but implementation is something that is totally private...05:45
ddaaA arch.backend object should not give the user access to all the crap behind. That's the reason because arch._arch is protected. The arch classes provide abstraction over the backend.05:46
ddaaTherefore, a arch.config which does not expose the actual backend internal would be more appropriate.05:47
=== ddaa has been round full-circle.
ddaanot... if I add a level of indirection :-)05:48
lifelessddaa: a) nothing in python is private, by definition. b) occams razor so applies here05:48
=== ddaa burns a candle Dijkstra
lifelessddaa: even then its still not private, then you are just playing namespace games to stop users getting at what they can get at anytime they want.05:49
ddaaSure, _protected is a convention. But conventions are operational.05:49
lifelesstomorrow, I won't make sense till then05:49
dilysMerge to rocketfuel@canonical.com/launchpad--devel--0: SourcepackageRelease.component foreign key fix, index for kiko, drop decoy BugInfestation table (patch-709)05:49
ddaaThe point being the backend configuration is public, not the backend guts is private to pyarch. I can access my own private parts.... Demeter.05:50
Kinnisonstub: is the 'foreign key fix' there the component fix we discussed earlier?05:50
=== Kinnison hugs stub
ddaalifeless: no need for you to make sense, I'm starting to make sense myself.05:51
Kinnisoncprov: gina exploded at me with a librarian error; I'm looking into it05:51
ddaaOne reason that issue is suddenly acute, is that there also the arch._tla.logger, the arch._tla.spawn which should not be there.05:51
Kinnisoncprov: also; sabdfl wants us to import hoary not warty to begin with :-)05:51
ddaaSo, with the command name, than makes three. Time to clean it up.05:52
=== debonzi|lunch is now known as debonzi
cprovKinnison: which librarian error ?06:16
KinnisonArgh; the librarian is returning an alias which isn't in the db06:16
cprovKinnison: hoary really looks more interesting to for dogfood and also quite easy to do 06:17
Kinnisoncprov: yeah; but until I work out why the librarian is returning aliases not in the db I'm not sure what to do06:17
cprovKinnison: uhm, wierd as the entire librarian behavior 06:17
spiv_Kinnison: Strange :/06:17
Kinnisonspiv_: aye; strange indeed :-(06:18
cprovspiv_: hi, dude ... how is it going ?06:18
Kinnisonspiv_: how can I get debugging info out of the librarian?06:18
=== Kinnison can currently reproduce the issue; so clues greatly appreciated
stubBy any chance are you getting the file id, not the alias id?06:22
spiv_Kinnison: Hmm, what sort of info do you need?06:22
spiv_Kinnison: It's probably already dumping various messages into twistd.log.06:23
spiv_Kinnison: If you're desperate, you can get it to open a port that you can telnet into and get a (pseudo) python prompt on the running process.06:24
spiv_(I think Twisted SVN may support sshing for that, too ;)06:24
Kinnisonstub: nup; the file id didn't come out either :-(06:25
Kinnisonspiv_: I'll look in twistd.log06:25
spiv_cprov: Pretty good.06:25
Kinnisonspiv_: Hmm, no twistd.log06:26
spiv_How did you start it?06:26
Kinnisontwistd -noy ...../server.tac06:26
stubalso, is it connecting to the correct database?06:26
Kinnisonyes; correct db06:26
spiv_Oh, -n... so it didn't fork.  In that case it logs to stdout.06:26
Kinnisonright; so I'm seeing no useful logging info then06:27
Kinnisonhow can I turn up the verbosity?06:27
Kinnisonseeing the sql would be a great boon06:27
spiv_vim a_file.py; (insert print statements)  ;)06:27
spiv_It's using SQLObject?06:27
=== spiv_ looks up the relevant flag.
stubturn on statement logging in postgresql06:28
KinnisonHow do I do that?06:28
KinnisonAnd where does it log to? and can I get to that place on zhongshan?06:29
stuboh - thought you were on a local box.06:29
Kinnisonnup :-(06:30
KinnisonI need an archive; and I don't have the bw to burn to mirror it to home06:30
spiv_Ugh, it's actually kinda messy accessing that with the initZopeless stuff.  Hmm...06:30
KinnisonCan I not set something simple like SQLBase.loglevel or something?06:31
stubthink so, but I don't remember the attribute to set. I'm sure it is documented somewhere ;)06:32
spiv_Kinnison: The simplest way atm is probably to edit sqlobject/dbconnection.py, line 28, and hard-code that line to True..06:32
spiv_Unfortunately, the sqlos/sqlbase layers get in the way of doing this nicely atm.06:33
spiv_Argh, in fact sqlos hard-codes it to 0.06:33
=== sabdfl [~mark@host217-37-231-28.in-addr.btopenworld.com] has left #launchpad []
spiv_So, edit lib/sqlos/adapter.py, line 46, and set that to True.  Ick.06:34
=== spiv_ looks into adding a debug flag to initZopeless
=== spiv_ is now known as spiv
KinnisonOkay; it's getting stuck on an insert failing06:38
KinnisonI think06:38
Kinnisonmaybe not06:39
=== Kinnison grrs; now it won't reproduce
spivKinnison: I'm about to merge andrew.bennetts@canonical.com/launchpad--devel--0--patch-223, it adds a debug flag to initZopeless.06:42
spivWhich will turn on the SQLObject debugging, i.e. sql statement logging.06:43
spivHopefully it will help you figure out what's going on, assuming you can reproduce it...06:45
KinnisonIrritatingly it's now running okay06:46
KinnisonI need to wait for it to fail again06:46
=== Kinnison mumbles something about heisenbugs
dilysMerge to rocketfuel@canonical.com/launchpad--devel--0: Add debug flag to initZopeless (patch-710)06:51
spivKinnison: If you merge that, you can change server.tac to say "initZopeless(debug=True)".06:53
Kinnisonspiv: once I can get the bastard to fail again I'll do that; thanks06:53
=== Kinnison has 1h30m to get things to break; fix the breakage; and get an import running on mawson :-)
Kinnisonstub: can you update mawson's db with that foreign key fix?06:55
stubThat would be naughty (running bits of the database upgrade without a code update), but I think we can get away with it in this case.06:56
=== Kinnison restarts this import in an attempt to reproduce the issue
spivstub: G'night.07:12
=== debonzi [~debonzi@] has joined #launchpad
KinnisonRight, I can reproduce it07:16
Kinnisonbut the sql log is just showing that it goes through the motions and doesn't commit the transaction for some reason07:16
KinnisonI think I know what it is07:19
=== Kinnison tries something
BradBcprov: ping07:22
KinnisonOkay; I can fix it if I cause addAlias to generate a transaction if it didn't have one07:25
=== ChanServ [ChanServ@services.] has joined #launchpad
=== lulu [~lu@host217-37-231-28.in-addr.btopenworld.com] has left #launchpad []
=== BradB is now known as BradB|lunch
Kinnisoncprov: ping?07:48
cprovKinnison: pong07:50
Kinnisoncprov: I have a few librarian fixes to finish and then I'll be in a position to get things going with a hoary import on mawson07:50
cprovKinnison: fantastic !07:51
cprovKinnison: do you need help ? 07:51
Kinnisoncprov: things will be fine as long as the katie db on mawson is ready07:51
cprovKinnison: fine07:55
=== Kinnison star-merges and runs a final test run
=== Kinnison is 1/3 of the way to knowing if his fix will do the trick
=== Kinnison bounces excitedly as gina imports more and more of warty
Kinnisonrah; it worked08:14
=== Kinnison commits
cprovKinnison: great08:14
=== Kinnison invokes the power of pqm-chinstrap-merge
=== Kinnison taps his fingers impatiently while pqm weaves its magic
dilysMerge to rocketfuel@canonical.com/launchpad--devel--0: librarian transaction fixes and gina build fix (patch-711)08:22
=== kiko [~kiko@] has joined #launchpad
Kinnisonhihi kiko08:24
=== Kinnison logs into mawson to do his bit
kikokiko the brave08:26
KinnisonWhat's the right way to create a distrorelease and distroarchrelease in the dogfood db?08:29
Kinnisondo I just go in and do it with SQL; or do we have a UI for it?08:29
kikothere is UI for that in Soyuz08:31
Kinnison"" A system error occurred.""08:32
KinnisonIf this can't be fixed in the next couple of minutes; I'm gonna hand-fiddle the db 'til I can run gina08:32
=== Kinnison has to go out in the next 5 minutes
Kinnisonor I can leave gina until later to start08:33
=== Kinnison can pop back here after he gets back from town
Kinnisonkiko, cprov: your call08:34
kikoKinnison, can you provide a traceback at least?08:34
Kinnisonkiko: how?08:34
kikoyour DB is cracked up somehow, btw08:34
kikousing the debug port?08:34
Kinnisondebug port? How do I get at that? this is launchpad on mawson08:35
kikoBradB|lunch, didn't elmo set us up to allow access to the debug port?08:36
kikoKinnison, there's a port listening HTTP that can fix that up08:36
=== Kinnison goes to get ready in the hope that you'll have an answer pretty soon
cprovKinnison: forget it and create distro and distrorelease by hand, please08:38
kikocprov, is creating a distro broken?08:38
cprovkiko: yes08:39
Kinnisoncprov: okay08:39
KinnisonMay I blat the warty distrorelease?08:39
KinnisonIt'll save time08:39
KinnisonI'll take that as 'yes'08:41
Kinnisonwhy the fuck are there things published to this distrorelease?08:42
Kinnisonwas that just random test data? can I blat it?08:42
=== carlos [~carlos@69.Red-80-33-181.pooles.rima-tde.net] has joined #launchpad
Kinnisonwell; since sabdfl wants me to blat publishing data each time; I guess I can...08:43
Kinnisonnoone should be using the publishing records outside of lucille/etc anyway08:43
Kinnisonlibpq.DatabaseError: FATAL:  database "katie" does not exist08:45
Kinnisonwho told me katie was ready?08:45
=== Kinnison imports the katie db
KinnisonGrr, the mirror lacks hoary properly08:52
KinnisonI can't do this now08:53
KinnisonMoan at elmo about how the tagfiles for hoary are empty08:53
=== Kinnison -> town
=== debonzi [~debonzi@] has joined #launchpad
=== debonzi is now known as lalo
=== lalo is now known as debonzi
!lilo:*! Hi all. If you're involved with accounting and/or bookkeeping for a small FOSS community (or other) not-for-profit entity, please consider maintaining a presence on #accounting .... support and consultation about accounting issues. Thanks!09:29
=== cprov [~cprov@] has joined #launchpad
=== BradB|lunch is now known as BradB
=== BradB reads (some) scrollback
BradBkiko: it does look like it. 9021 should be open, but isn't.10:18
=== carlos -> dinner
BradBkiko: when do you want to talk about what you guys can work on in malone btw?10:20
kikoBradB, I'll write to admins10:20
kikohum hum10:20
kikoI'm at a client today and tomorrow10:21
BradBit'll be a short conversation10:21
kikobut cprov and debonzi are happily hacking away till them10:21
kikomaybe tonight a bit later if I catch you on10:21

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!