pypy-sync developer meeting 27th October
Time & location: 1pm (30 minutes) at #pypy-sync
Armin Rigo (left early)
Eric van Riet Paap
Carl Friedrich Bolz
- activity reports (3 prepared lines of info).
All Attendees submitted activity reports (see `IRC-Log`_
at the end and 'LAST/NEXT/BLOCKERS' entries in particular)
- resolve conflicts/blockers: No blockers
Topics of the week
"this week on #pypy"
Michael volunteered to write a short "This week in pypy-land" text every week
(which would include the events in #pypy and other interesting stuff). This
would make e.g. navigating the #pypy-logs a lot easier but is also quite a lot
of work. Holger suggested that we should have a file in svn where everybody can
insert paragraphs. This was considered a good ideal. Michael plans to summarize
everything at the end of the week and send mails to pypy-dev.
Samuele brought up the issue of another making another release (0.8.0) soon
that highlights the translatable parser/compiler. He wants this because some
of the reportwriting he has to do is blocked by the fact that this release did
not happen yet. The features of the release would be working/translatable
astcompiler, translatable thunk object space, cleaned up translate_pypy and a
bit more speed.
The idea was to try make a small release that should not take too much work.
One of the biggest problems is that Holger is not really avaiable next week
which is when Samuele planned to do the release. Holger did a lot of the
infrastructure work for the last releases. Samuele noted that this release is a
bit different since we don't have new features that we want to rush in in the
last minute. It was decided that we should try to make a release friday next
week (November 4th, 2005), making a release branch on October, 28th. Over the
weekend the compliancy tests will be run to see whether compliancy got worse.
If this is the case we might have to reconsider. On Monday (October, 31th)
there will be a planning meeting. All in all there don't seem to be that many
people avaiable for the release: Samuele, Michael, Anders L., Arre(?), Carl
Friedrich (maybe a little), Christian (a bit).
gothenburg sprint topics
topic deferred to next pypy-sync.
Complete IRC log
**** BEGIN LOGGING AT Thu Oct 27 12:58:02 2005
Oct 27 12:58:02 * Now talking on #pypy-sync
Oct 27 12:58:18 * hpk (email@example.com) has joined #pypy-sync
Oct 27 13:00:29 cfbolz hi all!
Oct 27 13:00:34 hpk hi!
Oct 27 13:00:38 cfbolz it's 1 on my clock, so shall we start?
Oct 27 13:00:55 hpk fine for me. let's see if some others are awake :)
Oct 27 13:01:02 cfbolz :-)
Oct 27 13:01:02 pedronis hi, yes
Oct 27 13:01:05 mwh hello
Oct 27 13:01:06 arigo hi!
Oct 27 13:01:36 cfbolz ok, everybody post their activity reports!
Oct 27 13:01:39 ericvrp hi~
Oct 27 13:01:48 arigo LAST: holidays + reports + a stackless RPython interface
Oct 27 13:01:48 arigo NEXT: holidays + finish report draft-dynamic-etc + choose new laptop
Oct 27 13:01:48 arigo BLOCKERS: -
Oct 27 13:01:58 ericvrp Last: some genjs, mostly non-PyPy
Oct 27 13:01:59 ericvrp Next: dito
Oct 27 13:02:01 ericvrp Blockers: -
Oct 27 13:02:09 cfbolz LAST: statistic gathering, pypy unrelated
Oct 27 13:02:09 cfbolz NEXT: report
Oct 27 13:02:09 cfbolz BLOCKERS: none
Oct 27 13:02:27 hpk LAST: eu dissemniation/talk planning, codespeak admin. NEXT: pycon proposals, finalizing dissemination planning. BLOCKERS: somewhat non-permanent internet connection
Oct 27 13:02:46 pedronis LAST: some release prep, D04.4, looked at other reports, some other rep writing,rtyper level opts/code reductions
Oct 27 13:02:48 pedronis NEXT: release, finish D04.4, help with other reports
Oct 27 13:02:49 pedronis BLOCKERS: -
Oct 27 13:03:37 mwh last week: got a job! making and using refactoring tools
Oct 27 13:03:37 mwh next week: more of same
Oct 27 13:03:37 mwh blockers: -
Oct 27 13:04:14 mwh oh, and pycon proposals for next week
Oct 27 13:04:17 * aleale (firstname.lastname@example.org) has joined #pypy-sync
Oct 27 13:04:20 hpk bertf: ?
Oct 27 13:04:39 cfbolz aleale: your activity report?
Oct 27 13:04:42 aleale PREV: Writing report
Oct 27 13:04:54 aleale NEXT: WP9
Oct 27 13:05:00 aleale BLockers:-
Oct 27 13:05:11 bertf LAST: following svn activities NEXT: report BLOCKERS: -
Oct 27 13:05:31 cfbolz ok, let's proceed then to the regular topics
Oct 27 13:05:36 cfbolz there don't seem to be blockers
Oct 27 13:05:39 cfbolz "this week on #pypy"
Oct 27 13:05:39 cfbolz --------------------------
Oct 27 13:06:05 cfbolz michael volunteered to write something weekly about the ongoings of #pypy but would like to be notified about interesting stuff
Oct 27 13:06:38 hpk probably makes sense to have a weekly svn-commit file where people can insert paras?
Oct 27 13:07:01 mwh that makes sense so long as people actually do it :)
Oct 27 13:07:10 cfbolz indeed. and I fear they don't :-)
Oct 27 13:07:21 hpk well, all things need a bit of time to establish
Oct 27 13:07:25 mwh but yes, let's keep an activity log in svn
Oct 27 13:07:37 hpk but offering a structure makes it easier and point people to it
Oct 27 13:07:43 mwh and basically i'll summarize it and send it to pypy-dev each week
Oct 27 13:07:46 hpk should theey complain that a particular topic wasn't included
Oct 27 13:07:56 cfbolz makes sense
Oct 27 13:08:00 mwh i'll probably mostly maintain it as well to start with
Oct 27 13:08:42 * stakkars (email@example.com) has joined #pypy-sync
Oct 27 13:08:43 mwh extradoc/weekly or something?
Oct 27 13:08:51 cfbolz sounds good!
Oct 27 13:08:54 hpk mwh: yes, irc-weekly maybe
Oct 27 13:09:06 cfbolz does that approach make sense for everybody?
Oct 27 13:09:13 aleale yes
Oct 27 13:09:14 arigo yes
Oct 27 13:09:14 hpk yip
Oct 27 13:09:32 pedronis ok
Oct 27 13:09:37 mwh hpk: i was hoping to include more than what happens in irc..
Oct 27 13:09:46 ericvrp yeys
Oct 27 13:09:46 * arigo must leave, sorry
Oct 27 13:09:48 mwh though of course, that's currently the hardest thing to find
Oct 27 13:09:52 hpk mwh: ok, also ping me sometime so we make this more visible on the web page
Oct 27 13:09:58 * arigo has quit (Remote closed the connection)
Oct 27 13:10:04 mwh but, yes, agreed for now, let's move on
Oct 27 13:10:16 cfbolz good!
Oct 27 13:10:16 cfbolz "this week on #pypy"
Oct 27 13:10:16 cfbolz --------------------------
Oct 27 13:10:20 cfbolz sorry :-)
Oct 27 13:10:30 cfbolz new release
Oct 27 13:10:30 cfbolz --------------------------
Oct 27 13:10:30 stakkars should I post?
Oct 27 13:10:35 cfbolz stakkars: ok
Oct 27 13:11:03 stakkars DONE: benchmarks, some stackless optimizations
Oct 27 13:11:22 stakkars plus some windows stuff
Oct 27 13:11:24 stakkars NEXT: reviews, 2 weeks Beverly Hills
Oct 27 13:11:35 stakkars BLOCK: None (except for my mother's cancer)
Oct 27 13:11:53 cfbolz sorry for that
Oct 27 13:12:05 stakkars :( thx
Oct 27 13:12:11 cfbolz ok, on to the release!
Oct 27 13:12:28 pedronis I have started updating getting started
Oct 27 13:12:45 pedronis I can write the release announcement
Oct 27 13:13:07 pedronis the question is what to highlight, and when to freeze
Oct 27 13:13:21 cfbolz and how much work should go in
Oct 27 13:13:24 * hpk still needs to catch up regarding the release planning, apparently
Oct 27 13:13:42 cfbolz hpk: no, it's new to everybody :-)
Oct 27 13:13:49 hpk ah ok.
Oct 27 13:14:07 cfbolz basically this is the first big mention since paris
Oct 27 13:14:10 mwh the highlight of this release is the working/translatable astcompiler?
Oct 27 13:14:12 aleale I think speed and compiler are the highlights
Oct 27 13:14:19 hpk ok, so the idea is to do a 0.8 release and main features are - compiler integration and more speed?
Oct 27 13:14:22 pedronis well we have add quiite a few improvement since 0.7
Oct 27 13:14:39 stakkars include stackless?
Oct 27 13:14:41 pedronis but it also driven by EU things
Oct 27 13:14:43 cfbolz hpk: yes, and better translate_pypy, translatable thunks
Oct 27 13:14:49 hpk cfbolz: yip
Oct 27 13:14:57 mwh how well does "a flexible parser/compiler implementation" describe our situation?
Oct 27 13:15:05 mwh stakkars: that's phase 2 isn't it? :)
Oct 27 13:15:09 pedronis stakkars: that's a question, because stackless is not user visible
Oct 27 13:15:22 mwh and yes, no app level interface yet afaik
Oct 27 13:15:23 hpk i'd like to keep the release small and mean
Oct 27 13:15:27 hpk also to not incur too much work
Oct 27 13:15:30 cfbolz hpk: I agree
Oct 27 13:15:41 aleale hpk: I agree
Oct 27 13:15:42 pedronis hpk: the plan is to package what we have
Oct 27 13:15:43 stakkars it has no interface yet but by not crashing. But do we announce it as part of the release?
Oct 27 13:15:46 mwh i agree
Oct 27 13:15:58 * hpk wouldn't include stackless at this point
Oct 27 13:16:06 mwh what work needs to be done? who should do it?
Oct 27 13:16:06 pedronis we may need to disable some socket stuff
Oct 27 13:16:07 hpk but rather go for a 0.9 later in december (x-mas release) or so
Oct 27 13:16:13 pedronis which is very unstable
Oct 27 13:16:27 pedronis I mean failing tests
Oct 27 13:16:30 mwh i'm not amazingly busy, but need to be told what to do :)
Oct 27 13:16:52 pedronis mwh do you want to do more refactoring
Oct 27 13:16:54 hpk i guess i will need to be around during the release time so i'd actually first like to get an idea of the timing
Oct 27 13:16:57 mwh copy dist to release/0.8.x, disable socket on the branch, write an announce,ent?
Oct 27 13:16:59 * ericvrp would like to know when 0.8 is scheduled
Oct 27 13:17:01 mwh pedronis: yes
Oct 27 13:17:16 pedronis well, I think it should happen next week
Oct 27 13:17:22 hpk to state things upfront from my side: basically only 11th-15th if somewhat feasilbe as a release time, if i should be of much help.
Oct 27 13:17:39 mwh it would be good to get it done before then, imho
Oct 27 13:17:44 hpk pedronis: next week? uh.
Oct 27 13:17:50 pedronis hpk: yes
Oct 27 13:18:06 cfbolz how much is holger neede?
Oct 27 13:18:10 cfbolz needed
Oct 27 13:18:42 cfbolz or is everybody fine with 11-15?
Oct 27 13:18:49 stakkars maybe he can fork?
Oct 27 13:18:50 hpk why the rush? last times when i pushed for releases, we had like 3-
Oct 27 13:18:50 hpk i basically did much of the ground work the last times
Oct 27 13:19:16 * hpk thinks he _may_ be able to make it next week, is still in marocco though, but is a bit un-happy about the sudden rush
Oct 27 13:19:16 pedronis cfbolz: I need the release done
Oct 27 13:19:26 pedronis to finish the report
Oct 27 13:19:36 cfbolz pedronis: I see
Oct 27 13:19:57 pedronis and 11-15 I'm holiday myself (finally)
Oct 27 13:20:01 stakkars I'm only very little available until the 15th,at least.
Oct 27 13:20:04 cfbolz oh
Oct 27 13:20:27 cfbolz so basically next week michael, samuele and (maybe) me would be there. of the people here today
Oct 27 13:20:32 cfbolz aleale: what are your plans?
Oct 27 13:20:42 mwh how much work is a release?
Oct 27 13:20:47 aleale I am available
Oct 27 13:20:57 mwh if the answer is "not much" then ...
Oct 27 13:21:00 cfbolz mwh: depends on whether you want to look at compliancy?
Oct 27 13:21:00 pedronis hpk: it seems the part that needs you is uploading? or can that be done by someone else?
Oct 27 13:21:10 aleale mostly next week - less the week after
Oct 27 13:21:18 hpk mwh: it used to be quite a lot (some full 5 days from my side in the very last time)
Oct 27 13:21:29 hpk mwh: but i think it could be less now, i prepared some scripts and such by now
Oct 27 13:21:53 hpk pedronis: not quite, there are some details, have to think about it. but maybe you could all do it.
Oct 27 13:22:14 pedronis hpk: I also think is good think that releases don't depend on a single person
Oct 27 13:22:21 hpk pedronis: of course !
Oct 27 13:22:24 cfbolz :-)
Oct 27 13:22:29 hpk pedronis: i am pushing for that for a long time, mind you
Oct 27 13:22:36 cfbolz good training opportunity then
Oct 27 13:22:45 stakkars ping (checking if my GPRS modem is still online,train is moving)
Oct 27 13:22:49 aleale could we do a dry run in the weekend ?
Oct 27 13:23:04 hpk pedronis: and so far, my tries to get some ahead-planning for releases have met skepticism, haven't they?
Oct 27 13:23:06 mwh as in two days time?
Oct 27 13:23:13 mwh i am away climbing
Oct 27 13:23:31 pedronis mwh: when can we freeze
Oct 27 13:23:46 pedronis it seems it depends when you are happy with refactorings you are doing?
Oct 27 13:23:49 stakkars I can help on the weekend and Monday
Oct 27 13:23:57 cfbolz so do we try to get the release out, say, friday next week?
Oct 27 13:24:00 mwh pedronis: i can easily just stop for a few days
Oct 27 13:24:18 pedronis I would freeze tomorrow then and run the tests
Oct 27 13:24:31 cfbolz ok, so very fast :-)
Oct 27 13:24:44 mwh fine with me
Oct 27 13:24:49 pedronis well running the compliancy tests take some time
Oct 27 13:25:07 pedronis and we are in a situation when we can package what we have
Oct 27 13:25:08 stakkars why disable socket,toomuch missing or too many bugs?
Oct 27 13:25:17 pedronis failing tests
Oct 27 13:25:21 pedronis some even with segfaults
Oct 27 13:25:22 cfbolz segfaults
Oct 27 13:25:31 stakkars mupf
Oct 27 13:25:41 mwh platform dependent crap too
Oct 27 13:25:41 pedronis unless someone has a lot of time to put in that
Oct 27 13:25:47 pedronis is safer to disable
Oct 27 13:25:54 cfbolz I would say we disable
Oct 27 13:25:56 mwh pedronis: +1 to disable fro 0.8
Oct 27 13:26:00 pedronis is not really a release highlight and is far from finished
Oct 27 13:26:04 * hpk is not too happy about being mostly ignored
Oct 27 13:26:20 stakkars ?
Oct 27 13:27:17 pedronis hpk: our previous releases and always been charaterize by having to finish features at the last minute
Oct 27 13:27:23 pedronis s/and/have
Oct 27 13:27:57 * mwh points at the time
Oct 27 13:27:58 stakkars not so tjhis time if we infact throw out what isn't finished.
Oct 27 13:28:00 * ericvrp proposes to disable translator/js too
Oct 27 13:28:20 cfbolz hpk: you still there?
Oct 27 13:28:23 hpk yes
Oct 27 13:28:30 pedronis erivrp: yes, I tried to install js and lots of tests indeed fail
Oct 27 13:28:44 * hpk is thinking about if/how it all fits together.
Oct 27 13:29:07 cfbolz not at all :-(
Oct 27 13:29:14 stakkars what about llvm
Oct 27 13:29:18 hpk pedronis: the thing is, for example, that there is currently some migration regarding the codespeak backup underway
Oct 27 13:29:29 stakkars part of the release? disable the opaque stuff or finish?
Oct 27 13:29:48 cfbolz could we decide how to decide? some more decision at 2 on #pypy?
Oct 27 13:29:49 ericvrp stakkars: pypy-llvm works
Oct 27 13:29:50 hpk pedronis: i suggest to _try_ going for a friday release next week
Oct 27 13:30:09 pedronis hpk: that's ok
Oct 27 13:30:14 stakkars ericvrp: already? whow!
Oct 27 13:30:15 hpk pedronis: but merlinux also has another deadline on that very day :( (and the consortium meeting is there as well)
Oct 27 13:30:16 cfbolz ok, everybody fine with that?
Oct 27 13:30:27 aleale ok
Oct 27 13:30:32 stakkars ok
Oct 27 13:30:34 ericvrp ok
Oct 27 13:30:39 mwh ok
Oct 27 13:30:46 pedronis hpk: we should try to have a finalized tag before then
Oct 27 13:30:57 pedronis such that is a matter of pushing it out
Oct 27 13:31:04 hpk pedronis: of course. a ctually i sugest a planning meeting on monday or so
Oct 27 13:31:19 hpk pedronis: trust me, it is or at least used to be quite a lot of work to get a release out
Oct 27 13:31:26 pedronis hpk: makes sense
Oct 27 13:31:38 pedronis I still plan to make a branch tomorrow
Oct 27 13:31:42 cfbolz ok, let's close the meeting now
Oct 27 13:31:45 * hpk is off to another meeting
Oct 27 13:31:46 pedronis so that on monday we can see
Oct 27 13:31:50 cfbolz see you all later on #pypy!
Oct 27 13:31:52 cfbolz bye!
Oct 27 13:31:57 aleale bye
Oct 27 13:32:04 pedronis whether we have serious compliancy regressions
Oct 27 13:32:11 * aleale (firstname.lastname@example.org) has left #pypy-sync
Oct 27 13:32:16 * pedronis hopes not
Oct 27 13:32:20 ericvrp bye
Oct 27 13:32:25 mwh bye
Oct 27 13:32:28 * mwh (email@example.com) has left #pypy-sync ("ERC Version 5.0 (CVS) $Revision: 1.771 $ (IRC client for Emacs)")
Oct 27 13:32:34 * ericvrp (firstname.lastname@example.org) has left #pypy-sync
Oct 27 13:37:00 * bertf has quit ()
Oct 27 13:56:09 * stakkars has quit (Read error: 110 (Connection timed out))
**** ENDING LOGGING AT Thu Oct 27 14:03:31 2005