Open Source Meeting/2010-05-04

From Second Life Wiki
Jump to: navigation, search


Agenda Tuesday, 4 May 2010

  1. Weekly Snowglobe update - Merov Linden
  2. Your topic here!

If we run out of topics before the end of the meeting, we'll triage Snowglobe or Open Source specific Jira Issues from the list below.

Jira Issues

  1. Your issue here!

If you think an issue warrants a deeper discussion, please add it as topic to one of the agendas above.


[14:02] Merov Linden:
[14:03] Merov Linden: I saw that Boroondas edited that agenda
[14:03] Techwolf Lupindo: Has it been updated? :-0
[14:03] Merov Linden: looks nice :)
[14:03] Merov Linden: well, the overall template
[14:04] Techwolf Lupindo: Now it just needs instructions on how to add an issue or jira number.
[14:04] Merov Linden: so, first order of work is my update
[14:04] WolfPup Lowenhar: i wonder how many saw my patch to the way chat and IM logs are saved ?
[14:04] Techwolf Lupindo: I got svn 1.4 branch building on my systme
[14:04] Merov Linden: cool Techwolf!
[14:05] WolfPup Lowenhar: im doing V2 branch
[14:05] Merov Linden: I've been working on several things on SG2.0 this past few days
[14:05] Merov Linden: first export scripts:
[14:05] WolfPup Lowenhar: after i got to finaly build on my system
[14:05] Aleric Inglewood: I still got 1.4 .. did it ever not compile?
[14:06] Techwolf Lupindo: We had broken/missing files from S3. That got fixed.
[14:06] Merov Linden: there was a problem with eol sometimes and I've been trying to fix it, unfortunately breaking the build (and hence the export) in the process
[14:06] WolfPup Lowenhar: im having to build v2 on vs2005 though still working on getting it to build in vs2008 or vs 2010
[14:06] Merov Linden: I've been working on this this morning, hope it'll be fixed by EOD
[14:06] Aleric Inglewood: You're not compiling standalone Techwolf?
[14:07] Techwolf Lupindo: Aleric, yes, standlone, but the artwork and libs disapeared off of S3.
[14:07] Merov Linden: another news on 2.0 is that we now completed the creation internally of viewer-public
[14:07] Merov Linden: so I'll be hooking the export directly to this asap
[14:07] Merov Linden: (as soon as I fix the export script...)
[14:07] Techwolf Lupindo: Will there be commint comments?
[14:07] Merov Linden: that should be pretty cool!
[14:08] Merov Linden: Techwolf: yes!
[14:08] Merov Linden: I have a patch in the script to export the comments in clear
[14:08] Techwolf Lupindo: Good. I'me sure a lot of tpv would like to cherry pick some bug fixes for backporting.
[14:08] Merov Linden: yup
[14:08] Merov Linden: The first sync though will be pretty bulky again
[14:09] Merov Linden: but that should become less of a problem once we do that automatically
[14:09] Aleric Inglewood: Did some port VWR-12984 to 2.0, or will someone? I'm not using 2.0, but that patch should be added to all viewer, also snowglobe 2.0, and also viewer 2.0.
[14:09] JIRA-helper:
[#VWR-12984] 1.23: Water flickers and disappears in patches
[14:09] WolfPup Lowenhar: so what dose any one here think of SNOW-622
[14:09] JIRA-helper:
[#SNOW-622] Adding Date Stamp to Chat and IM logs
[14:10] Merov Linden: Aleric: yes, that should be ported. If you could do it, that'd be great.
[14:10] Merov Linden: I don't know how different the code you modified is in 2.0
[14:10] Aleric Inglewood: I don't use 2.0 ... didn't even download it yet :/
[14:11] WolfPup Lowenhar: well im currently runing on a build with that in it
[14:11] Merov Linden: I alternate between the 2 all the time
[14:11] Aleric Inglewood: Maybe Thickbrick is willing to do the port. He's co-author.
[14:11] Merov Linden: Also, I've been starting going systematically through all the 1.x commits and see what needs to be ported
[14:12] WolfPup Lowenhar: merov it adds the ability to date stamp the chat and IM logs if your saving them
[14:12] Merov Linden: that's going to keep me busy for a while (few days) for sure
[14:12] Merov Linden: lets talk about that 622 from WolfPup
[14:12] Merov Linden: :)
[14:12] Aleric Inglewood: Error
[14:13] Aleric Inglewood: err
[14:13] Aleric Inglewood: nm - wrong number :)
[14:14] Aleric Inglewood: Oh, I looked at that before and I'd not like to use it. So, it should be optional :p
[14:14] Free Portal: snow-622 is Split files by timesteamp?
[14:14] Merov Linden: Henri has the same comment
[14:14] Merov Linden: and I think it's fair
[14:14] WolfPup Lowenhar: the second patch i added has a switch in advanced>debug so that you can turn it on or off
[14:15] Merov Linden: /me checks that patch
[14:15] Aleric Inglewood: Seems rather pointless to me - except and ONLY if you want to use log rotation.
[14:15] WolfPup Lowenhar: free actualy by date stamp
[14:15] Aleric Inglewood: And I don't, I want to keep ALL logs, forever.
[14:15] WolfPup Lowenhar: aleric then you do not mind having to search through megs od text to find what you need if nessary?
[14:16] WolfPup Lowenhar: this would give yu the ability to look at a particular days cat logs
[14:16] Techwolf Lupindo: For log rotation, just logrotate, it has a new feature to rotate by date instead of just incrmenting a number.
[14:17] Merov Linden: I always cringe when I see date formatting being done "manually"...
[14:17] Merov Linden: we should use ICU in the viewer
[14:17] Aleric Inglewood: Wolf: What difference does it make in that case whether you have 1000 files of 1 Meg, or 1 file of 1000 Megs? Same number of megs.
[14:17] Merov Linden: for all those dates and time formatting, leaving them locale sensitive
[14:18] WolfPup Lowenhar: actualy the dat stamp seems to be working off of UTC and not local
[14:18] Aleric Inglewood: So far I had no problems finding anything back. I surely can't remember the date on which something was said - at most I can remember who said it, or that it was in local chat.
[14:18] Boroondas Gupte: locale sensitive might not always be the best choice
[14:18] Techwolf Lupindo: The date/time sytem with the SL codebase is severlly borken.
[14:18] Merov Linden: Techwolf: +1
[14:19] Merov Linden: that was a general comment indeed
[14:19] WolfPup Lowenhar: it is getting utc time
[14:19] Aleric Inglewood: WolfPup: Obviously, I search not with my eyes :p ... I use 'grep'. And I have time stamps in the logs, so if I know a date range then I add that to the regular expression.
[14:20] Boroondas Gupte: Another concern: Chat and IM windows use to show a short backlog of previous conversations. I guess they take that from the logs. If the last conversation would be from a previous day/month that'd fail, wouldn't it?
[14:20] Techwolf Lupindo: There needs to be a project in witch that goes through the codebase and convert everything to UTC
[14:20] WolfPup Lowenhar: aleric to you host regular daily events?
[14:20] WolfPup Lowenhar: with lots going on in local caht?
[14:21] Merov Linden: well, OK with me to have this extra printout as an option
[14:21] Merov Linden: we need some more code review though
[14:21] Merov Linden: Someone to take the lead on that one?
[14:21] Merov Linden: /me realizes that only Aleric and him are committers here...
[14:22] Latha Serevi: btw, MichelleZ is en route now.
[14:22] Merov Linden: good :)
[14:22] Techwolf Lupindo: Its a usfull obsure featrue for some folks. Keep at it wolfpup and imporov it to account for international settings and keep it a default off debug setting. Not many would use it.
[14:22] WolfPup Lowenhar: i did try to use the same date/time system that it down in the timestamp setion of lllogchat but it kept failing to compile or would not give the right output to the filename till i hard coded it
[14:23] Merov Linden: K, I'm going to set "Fix version" to "2.0" so we pick it up and it shows up on the committers' radar
[14:24] Techwolf Lupindo: wolfpup, you may have to add antoher debug setting to use UTC or local time and figure out a way to pick up the date/time stamp format of the local user.
[14:24] Aleric Inglewood: I don't mind when it's implemented as optional of course, but I'd be upset if it ever was (suddenly) turned on, alpha/beta or debug. It would screw up my logging. It should be off by default and at any point in the commits.
[14:24] WolfPup Lowenhar: when i was first trying to get it working i would get this --- chat-year,datetime,slt-mthnum,datetime,slt-day,datetime,slt as the output filename
[14:25] Boroondas Gupte: literally? or are those placeholders?
[14:25] WolfPup Lowenhar: even though the patch that includes the debug setting it defaulted to on it could be set to be defaulted to off
[14:25] Merov Linden: /me edited the JIRA
[14:26] WolfPup Lowenhar: that is exactly the file name i would get when i first tried
[14:26] Aleric Inglewood: Just use some new persisting debug setting that is off by default
[14:26] Merov Linden: ok, one thing I wanted to review is the list of patches assigned to 1.4
[14:26] WolfPup Lowenhar: the is what the debug setting is for
[14:27] WolfPup Lowenhar: it is just when i made the patch file i ad it defaulted to on
[14:27] Techwolf Lupindo: wolfpup, rework the patch to default off and put all the intro comments at the top of the patch where they can be ignored by the patch program. No need to have all those comments in the code. Just a one liner explaing the debug setting/feature is enought.
[14:27] Aleric Inglewood: WolfPup, yes but
[14:27] Aleric Inglewood: + <key>Value</key>
+ <integer>1</integer>
[14:27] Aleric Inglewood: should be <integer>0</integer>
[14:28] Techwolf Lupindo: [17:25] <MichelleZ> ping t_brick or Techwolf if either of you are inworld at the meeting i can't log in for various silly reasons ;-/
[14:28] Aleric Inglewood: Also move the '//static' to where it belongs
[14:29] WolfPup Lowenhar: ok ill make a new patch file and up load it
[14:29] Boroondas Gupte: :-)
[14:29] Merov Linden: /me agrees with Aleric's quick review
[14:29] WolfPup Lowenhar: actualy that static was already there
[14:29] WolfPup Lowenhar: i just add the coments to it is all
[14:30] Merov Linden: k, we'll check your patch WolfPup :)
[14:31] Merov Linden: I'd like to review some 1.4 pending patches
[14:31] Aleric Inglewood: There are more style things... like the declaration of all new local variables should be inside the scope of the new if().
[14:31] WolfPup Lowenhar: ok ill let you know as soon ad i upload a new patch with the defaut as off
[14:31] Merov Linden: first would be: SNOW-440
[14:31] JIRA-helper:
[#SNOW-440] Media plugins have no way to log errors.
[14:33] Techwolf Lupindo: /me checks "Hmm...I have three patches waiting, one of them is for 1.4."
[14:33] Merov Linden: on SNOW-440, I think Thickbrick should go ahead and commit
[14:34] Merov Linden: that's one we should take in SG2.0 BTW, debugging plugins is hit and miss without some trace
[14:34] Techwolf Lupindo: +1 on commiting that patch. I get media errors sometime and nothing on the console.
[14:34] Merov Linden: Techwolf: please, go ahead with your patches
[14:35] Merov Linden: type the JIRA here and we'll go through them
[14:35] Aleric Inglewood: I read 440, and I think it won't help much... at least not in the cases of core dumps.
[14:35] Aleric Inglewood: So, then you got to ask yourself if there isn't a better way.
[14:35] Techwolf Lupindo: /me picks the most usfull one first....
[14:35] Techwolf Lupindo:
[14:35] JIRA-helper: [#SNOW-630] Finished port of SNOW-307
[14:36] Merov Linden: Aleric: true, but too often, the only info we have from plugin issues is "won't work!", no crash, no trace, nothing...
[14:36] Techwolf Lupindo: Right now, the 2.0 code base is using the same mine_type file for all three OSes.
[14:36] Merov Linden: nothing in the log in particular
[14:36] Techwolf Lupindo: /me waits.
[14:37] Boroondas Gupte: in doc/contributions.txt, also add snow-307, so we know it's been ported
[14:37] JIRA-helper:
[#SNOW-307] Linux viewer builds renames gstreamer plugin to quicktime.
[14:37] WolfPup Lowenhar: ok new patch with default set to off added
[14:37] Merov Linden: ah right
[14:37] Merov Linden: this was so fuggly...
[14:38] Aleric Inglewood: Ok, it would help if you get that info from users. For developers, I think it should be fully supported and possible to run the SLPlugin in a debugger as the only right way to debug it. And THAT has a problem, since right now any normal output to stdout (I think that this is the reason anyway) totally confuses the communication between parent process and the plugin with as the normal result "nothing" and "it doesn't work".
[14:38] Aleric Inglewood: What we need is that the pipes between the parent and SLPlugin doesn't use file descriptors 0, 1 or 2 anymore. But use 3 and 4 for example
[14:40] Merov Linden: You're certainly right Aleric
[14:40] Merov Linden: Debugging plugins is an issue that needs a solution
[14:40] Merov Linden: We should get Monroe involves in such a discussion as he's the keeper of that code
[14:41] Merov Linden: I wonder what they use to debug their plugins...
[14:41] WolfPup Lowenhar: on my system i can not build the to have debug available
[14:42] Merov Linden: changing this though should be done asap, before we have too much of a legacy of existing plugins
[14:42] Merov Linden: especially 3rd party plugins
[14:42] WolfPup Lowenhar: had to set VS2005 to Release only not even Releasewith Debug
[14:43] Techwolf Lupindo: Boroondas, re:snow-307, where sould that be place in contrub doc? I notice a Linden did the patch for it.
[14:43] Merov Linden: Stange WolfPup
[14:43] Merov Linden: which Windows OS flavor do you run?
[14:43] Boroondas Gupte: oh
[14:44] Boroondas Gupte: um ... then we need a better mechanism to track what ports are in SVN
[14:45] Techwolf Lupindo: re:snow-307. I think I put it under me as I was the ornigle bug reporter.
[14:45] Merov Linden: there's no problem having Linden contributing patches... they just don't get the credit in the contrib doc
[14:45] Merov Linden: but you'll get the redit for 630 Techwolf :)
[14:46] Boroondas Gupte: yeah, but for patches ported over from 1.x to 2.x, we used to add both issue numbers, so that a diff could show what still needs porting
[14:46] Boroondas Gupte: if the original isn't in contributions.txt, because it's a linden patch, that approach won't work
[14:46] Merov Linden: yes... hmm...
[14:46] Techwolf Lupindo: Boroondas, I've did that in
[14:46] JIRA-helper: [#SNOW-617] Missing contributions.txt additions.
[14:47] Merov Linden: well, we can't go back to add those
[14:47] WolfPup Lowenhar: windows 7 32-bit
[14:48] Merov Linden: Still Boroondas, how having the original bug in contrib helps? we're not changing things for those are we?
[14:49] Merov Linden: for instance: 630 being done would change something on 307 line?
[14:49] Merov Linden: (assuming it was there)
[14:49] Techwolf Lupindo: Merov, snow-307 was comminted with no doc entry.
[14:49] Merov Linden: how do you see the association 630/307 in the diff?
[14:50] Boroondas Gupte: nope. lines present in the 1.x contribution.txt but missing from the 2.x contributions.txt would indicate issues that still need porting
[14:51] Merov Linden: oooo... you would take 307 out then
[14:51] Merov Linden: right?
[14:51] Boroondas Gupte: hu?
[14:51] Techwolf Lupindo: snow-307 is missing from the 1.x txt. I just happen to noticed it got borke in the 2.0 codebase.
[14:52] Merov Linden: I know Techwolf, I'm trying to understand that tracking port idea and I don't think it works
[14:52] Merov Linden: let's *assume* for a sec that 307 was there (contributed by Aimee for instance)
[14:53] Merov Linden: under the proposed tracking, we would take that attribution out (so it shows as "missing") when committing the 630 port
[14:53] Boroondas Gupte: no
[14:54] Merov Linden: ok, then I didn't understand
[14:54] Boroondas Gupte: in branches forked before it was commited it would already be missing
[14:54] Boroondas Gupte: in branches forked afterwards, the entry AND the patch would be present
[14:54] Boroondas Gupte: when you port it to a branch where it was missing, you should also add the "old" number, so it doesn't show as missing anymore
[14:55] Techwolf Lupindo: Its just snow-307 got commneted with entry, so the diff would not show it. But for everything else, the diff works for checking the ports.
[14:55] Techwolf Lupindo: That is untill everything got ported over then this would be a non-issues at that point.
[14:55] Boroondas Gupte: indeed
[14:56] Merov Linden: but, we have everything in the current 2.0 contrib.doc, including ref to patches that need to be ported
[14:56] Techwolf Lupindo: Boroondas is trying to throw a monkey wrench into the works. :-)
[14:57] Aleric Inglewood: I don't even follow the problem :)
[14:58] Techwolf Lupindo: The problem is snow-307 is missing from contrube.txt and broondas wants it added somewhere.
[14:58] Aleric Inglewood: Just commit it
[14:58] Boroondas Gupte: yeah
[14:58] Techwolf Lupindo: +1 Aleric
[14:59] Boroondas Gupte: I don't want to add it anywhere, I just noticed our previous approach didn't work.
[14:59] Merov Linden: who decided on this approach?
[14:59] Techwolf Lupindo: We wasted enought time and there is two more patches.
[14:59] Merov Linden: I'm not even aware of it...
[15:00] Aleric Inglewood: If the problem is "how to track if a patch that was committed to SNOW went into the internal branch" ... then I hope that works automatically now, ... we are using a repository for that no? *confused*
[15:00] Boroondas Gupte: IIRC I suggested it some time ago and you (Merov) approved it.
[15:00] Merov Linden: /me as confused as Aleric...
[15:00] Techwolf Lupindo: /me is taking a self timeout, afk
[15:00] Boroondas Gupte: But I don't think there was any formal decision on it
[15:00] Merov Linden: no, I approved that people contributing ports would log a JIRA and get credit for a port
[15:01] Boroondas Gupte: anyway, back to JIRAs
[15:01] Aleric Inglewood: What I am "afraid" of is that when an internal dev ports a snowglobe patch to the internal repro, that they'd do that manually - changing bits - perhaps forgetting the contributions.txt hunk etc - such that all relationship between the two vanish.
[15:01] Merov Linden: I may have misunderstood the original discussion but I never though it'd be used as a port tracking mechanism
[15:02] Aleric Inglewood: In the case of 'git' and a prefect copy of the patch, it would just note that it was already applied to SG when merging it back and there would be no problems. But that only works if the internal devs take out patches literally - and only then maybe add changes on top of it in a new commit
[15:02] Aleric Inglewood: take our*
[15:03] Merov Linden: ok, before we review the 2 other patches from Techwolf, lets agree to move that tracking port discussion to opensource-dev
[15:03] Boroondas Gupte: ok
[15:04] Merov Linden: it's a good idea, I'm just not convinced the solution proposed works (and I'm unclear on what the solution is)
[15:04] Merov Linden: /me edits 630
[15:04] Techwolf Lupindo: One for 1.4 codebase,
[15:04] JIRA-helper: [#SNOW-497] Add option to run unit tests for STANDALONE
[15:04] Aleric Inglewood: /me plans on stealing Latha's stuffed animal toy.
[15:05] Techwolf Lupindo: /me chuckles as he was allready plotting ongetting that toy...
[15:07] Techwolf Lupindo: No commnets, patch must be that good. ;)
[15:07] Merov Linden: 497 is good, what about porting it to 2.0 too?
[15:07] Boroondas Gupte: dunno if that's related, but I think I still can't build with LL_TESTS set to ON
[15:07] Techwolf Lupindo: No port needed.
[15:07] Techwolf Lupindo: LL_TESTS is in 2.0 codebase.
[15:07] Merov Linden: kk
[15:07] Boroondas Gupte: ah, right
[15:08] Techwolf Lupindo: I created that patch before 2.0 was dropped and then updated it after 2.0 was dropped to use open source folks.
[15:08] Techwolf Lupindo: use=us
[15:08] Merov Linden: k, edited the JIRA
[15:08] Merov Linden: next and last from you Techwolf?
[15:09] Techwolf Lupindo: An easy one.
[15:09] JIRA-helper: [#SNOW-617] Missing contributions.txt additions.
[15:09] Merov Linden: ah right, yes, go for it!
[15:09] Techwolf Lupindo: I"me not a commiter. :-(
[15:10] Merov Linden: ooops... k, I'll take it...
[15:10] Merov Linden: /me thinks we should consider making Techwolf a committer
[15:10] Aleric Inglewood: What is your largest patch TechWolf?
[15:10] Aleric Inglewood: let me look how it looks :)
[15:11] Techwolf Lupindo: /me thinks
[15:11] WolfPup Lowenhar: merov i have added an updated patch to SNOW-622 with default set to off
[15:11] JIRA-helper:
[#SNOW-622] Adding Date Stamp to Chat and IM logs
[15:11] Merov Linden: Thanks WolfPup!
[15:12] Merov Linden: please bring the issue to opensource dev so that others (not present here) have a chance to review
[15:12] WolfPup Lowenhar: ok
[15:12] Merov Linden: thanks!
[15:12] Merov Linden: ok, 3:12 already!
[15:12] Boroondas Gupte: WolfPup, some of the mentioned issues, remain: //static belongs after the new comment. The initializations belong into the if scope, credits don't belong into the in-code comment
[15:13] Techwolf Lupindo: I just did a jira search. None of them stand out a large ones except if you count all the gcc warning fixes as one.
[15:13] Merov Linden: need to check that build before I leave so I need to go now :)
[15:13] Merov Linden: see you Thursday
[15:13] Boroondas Gupte: cu
[15:13] Nicky Perian: bye
[15:14] Techwolf Lupindo: I am planning on making a large one for WVR to fix all the linux packaging issues that have been commited to snowglobe 2.0 so they can import it internially.
[15:14] Boroondas Gupte: well, that'll just be a rediff of a series of existing patches, right?
[15:16] Techwolf Lupindo: yes, plus one fix for the viewer-internail code base. They left out a chunk in the
[15:16] Aleric Inglewood: WolfPup: writing patches works as follows: you put heaps of blood sweat and free time into something, then you only get negative feedback; If that puts you off and you give up because "it won't be accepted anyway", then it won't ever be accepted. However, that feeling is not how you should feel. The correct feeling is to try and take EVERY comment and fix it to satisfy it (or discuss it first if you really disagree). Then come with a patch that addresses each and every comment - perhaps you get again comments, if so, you do it again. That way the patch will reach a point were suddenly people will say: ok, commit that. That's how it works ;)
[15:17] Techwolf Lupindo: Aleric, the biggest one that I can find right off paw is
[15:17] JIRA-helper:
[#SNOW-517] viewer manifest is incomplete on 64bit
[15:17] Aleric Inglewood: The worst thing that can happen (happened to me) if when people have no time to review it: so you get no comments, but also no commit :/
[15:17] Aleric Inglewood: As long as you get comments, you have progress.
[15:18] Boroondas Gupte: /me sings: Always look at the bright side ... of patchs! Di-dumm, di-dumm-di-dumm-di-dumm!
[15:18] Aleric Inglewood: techwolf: that isn't even C++ ... I can't judge based on that if you'd be a good committer :/
[15:19] Aleric Inglewood: Social life calls... cya next week I guess.
[15:19] Techwolf Lupindo: /me shrugs "Sorry."
[15:20] Boroondas Gupte: I think more important for a commiter than whether he/she writes good code, is whether they can judge others' code
[15:20] Aleric Inglewood: Certainly for a reviewer - which are very important (shortage).
[15:21] Boroondas Gupte: (those two might correlate, of course)
[15:21] Aleric Inglewood: But reviewing is harder than writing a patch yourself imho
[15:21] Boroondas Gupte: indeed
[15:21] Aleric Inglewood: I'd be a very bad reviewer I'm afraid :/
[15:22] Aleric Inglewood: My TP arrived
[15:22] Aleric Inglewood: bye!
[15:22] Boroondas Gupte: and your code is kinda hard to review
[15:22] Techwolf Lupindo: see you
[15:22] Boroondas Gupte: bye

Generated with SLog Wikifier