Open Source Meeting/2008-10-02

< Open Source Meeting

Agenda

Have spotted that DBUS has been added to the viewer in trunk, any news on this and plans for IPC? (Michelle2 Zenovka)

Transcript

[14:02] Rob Linden: hi folks
[14:02] Michelle2 Zenovka: Hi ya
[14:04] Rob Linden: hmm....wiki seems a little sluggish still
[14:05] Michelle2 Zenovka: IT WAS TIMING OUT ON ME EARLIER
[14:05] Rob Linden: oh, grrr....hit by the problem with the Open Source Meeting page
[14:05] Rob Linden: https://wiki.secondlife.com/wiki/Open_Source_Meeting/Agenda
[14:05] Rob Linden: the parent page still has that template since I haven't gotten aroud to reverting it
[14:06] Rob Linden: anyway....let's get started
[14:06] Rob Linden: Have spotted that DBUS has been added to the viewer in trunk, any news on this and plans for IPC? (Michelle2 Zenovka)
[14:06] Rob Linden: well....
[14:07] Rob Linden: Tofu added that in order to specifically deal with URL passing, iirc
[14:08] Michelle2 Zenovka: Thats better
[14:08] Rob Linden: I don't know that there's much I can say about it beyond that, but it's a worthy topic for the mailing list
[14:08] Rob Linden: Tofu generally doesn't show up for these meetings because this isn't a great time for him
[14:08] Michelle2 Zenovka: Sorry missed last few mins can some one catch me up
[14:08] Q Linden: Sorry i'm late
[14:09] Khyota Wulluf: wow, url passing works now
[14:09] Rob Linden: we were late getting started in general
[14:09] Rob Linden: Khyota: what platform/viewer version are you on?
[14:09] Soft Linden: Yes, that was only added for URL passing. At present, that's the only intended use.
[14:10] Khyota Wulluf: i am on linux 64 using the latest openal branch
[14:11] Rob Linden: oh, cool....I'm guessing the dbus stuff is there, too
[14:11] Rob Linden: loves to hear early reports of nefarious schemes going according to plan
[14:12] Soft Linden: grins.
[14:12] Michelle2 Zenovka: Is it building ok now or are you still having issues finding the headers
[14:12] Michelle2 Zenovka: the dbus ones?
[14:12] Khyota Wulluf: well i had to move them though
[14:12] Khyota Wulluf: copy*
[14:12] Michelle2 Zenovka: Looks like there is an issue with the cmake rules finding the dbus headers in standalone
[14:13] Soft Linden: If it's just a standalone cmake issue, I'd strongly encourage a submitter patch. We might not get to a fix quickly on that
[14:14] Michelle2 Zenovka: Khyota, can you open a JIRA on that and ping me the number, i can test/patch as soon as i have a few minutes
[14:14] Rob Linden: put it in "VWR" marked with a "Source Code" component
[14:14] Khyota Wulluf: sure thing
[14:14] Rob Linden: thanks Khyota!
[14:15] Rob Linden: this group of people here keeps a closer eye on the "Source Code" component than other pockets of jira
[14:16] Rob Linden: if you look at the volume of email that the "jira-notify" email list gets, you get a sense of why we aren't keeping an eye on every submission
[14:16] Rob Linden: ....though Alexa (all hail!) does a pretty good job of monitorig that traffic these days
[14:17] Rob Linden: for those of you that don't know what I'm talking about: https://lists.secondlife.com/cgi-bin/mailman/listinfo/jira-notify
[14:17] Rob Linden: annnyway
[14:18] Rob Linden: anything else on the dbus/ipc topic?
[14:18] Michelle2 Zenovka: yea i'm not yet on that
[14:18] Khyota Wulluf: trying to decide what to label the issue
[14:19] Michelle2 Zenovka: Can't find dbus.h in linux standalone build... etc ;-)
[14:19] Rob Linden: "DBus libraries hardcoded to single location" maybe?
[14:19] Michelle2 Zenovka: it does not appear to be hardcoded
[14:19] Michelle2 Zenovka: but something in the rules is not quite right
[14:19] Khyota Wulluf: DBus librarys not found during compile?
[14:19] Rob Linden: yeah, that works.
[14:20] Khyota Wulluf: ok good
[14:20] Rob Linden: please reference this conversation so that Tofu knows to ping us about this if he has questions
[14:20] Michelle2 Zenovka: hopefully i can test and patch quickly as well
[14:21] Rob Linden: any other dbus/ipc things we should talk about here?
[14:21] Soft Linden: Is there anything else we really should be using dbus for which we aren't? What did you hope to see, M2?
[14:22] Soft Linden: Or only a surprise that it was there at all?
[14:22] Rob Linden: I should mention one of the things Tofu brought up which is really vaguely related.....
[14:24] Michelle2 Zenovka: I was just curious as i spotted it appearing in trunk
[14:24] Rob Linden: he's also been doing some work on GStreamer, and has hit an upstream problem where GStreamer blocks the main loop when doing ??? (/me tries to remember....mime type lookup?)
[14:26] Rob Linden: GStreamer does have a DBUS interface, so we *could* conceivably move it out to its own process that way. not likely we'd solve that problem with dbus, but not 100% out of the question
[14:26] Michelle2 Zenovka: hmm nice
[14:26] Soft Linden: Does dbus use shared memory?
[14:26] Michelle2 Zenovka: i love these subitle things that sneak up and bite you on the ass
[14:27] Soft Linden: I just ask because a raw decode over a pipe could be very bus unfriendly :)
[14:28] Dark Brown: Coyote head with sculpties: Squirrel Wood grabs a newspaper and smacks Khyota Wulluf across the face!
[14:28] Rob Linden: oh, Soft, that's probably not how it'd work....it'd be more like how we do voice today (relying on the underlying environment to do mixing)
[14:28] Squirrel Wood: cuda?
[14:28] Khyota Wulluf: o.o
[14:28] Soft Linden: Right, but we still need the decoded video to blast back into texture memory. It's not like vivox where it can just set up its own audio channel.
[14:29] Rob Linden: ah....good point
[14:29] Soft Linden: Curious what's there though. Will look, ask Tofu about it if it looks like it makes sense
[14:30] Soft Linden: If we could feasibly get the decode out of our app space, it could help with threading, and with avoiding another vulnerability like the Quicktime one.
[14:30] Soft Linden: Can't smash a stack you don't have!
[14:31] Rob Linden: well, the vulnerability would still be there as long as Quicktime/whatever wasn't fully sandboxed by the OS
[14:31] Soft Linden: Right, but that's what dbus buys you. Interprocess communication. It couldn't trivially hijack the SL process.
[14:31] Soft Linden: I don't think even Windows grants raw access to other processes' address space.
[14:32] Rob Linden: yeah, but full access to the filesystem and system calls is enough to do a lot of damage
[14:32] Soft Linden: Truth
[14:33] Michelle2 Zenovka: although on linux it would be trivial to chroot
[14:33] Michelle2 Zenovka: if you were parinoid
[14:33] Rob Linden: yeah, once you actually start doing things like jailing plugins, then it actually starts getting interesting from a security perspective
[14:34] Soft Linden: Oh bah - that was our only agenda item?
[14:34] Rob Linden: yeah, that's why I'm letting it ramble ;-)
[14:34] Soft Linden: :)
[14:34] Mm Alder: I'd like to ask a question about puppeteering. OK?
[14:34] Michelle2 Zenovka: I was going to bring up another but it was a bit borderline on topic and seems to have been grabbed
[14:34] Rob Linden: go for it
[14:35] Rob Linden: Mm: what about puppeteering?
[14:35] Mm Alder: Do you know of anyone working with the code?
[14:36] Rob Linden: well, yeah, there's some fiddling going on from what I understand. Linden Lab has put it on ice, and it represents a bit of an evolutionary dead end
[14:36] Aimee Trescothick: were you are the orange island session a couple of hours ago?
[14:37] Mm Alder: The code is off of an old branch. Is it worth bringing it up to the current branch?
[14:37] Aimee Trescothick: there were a few people talking there that are working off it
[14:37] Mm Alder: Aimee: Missed that. Didfn't hear about it. Any refs you have?
[14:38] Aimee Trescothick: I'll dig back through my history a mo, I suspect a transcript may pop up somewhere
[14:38] Rob Linden: hmmm.....I don't know if anyone has cycles to take in a patch that updates that, but I can see. if someone with a contrib agreement wanted to work on it in the sandbox area on svn.secondlife.com, I think we could set that up
[14:38] Aimee Trescothick: it was being chaired by Dusan Writer, so he might be a good person to ping about it
[14:39] Mm Alder: I'm interested in using it as a way to input animations, but I don't want to get too far unless I can expect it would get into the mainline.
[14:39] Rob Linden: hmm.....
[14:40] Aimee Trescothick: Roger Fullstop was the main guy that was talking about working with it
[14:40] Mm Alder: Thanks Aimee. Will do.
[14:40] Rob Linden: there *might* be a sensible way to subset that code
[14:40] Rob Linden: er.....not sure why I verbed the word "subset" there :)
[14:41] Mm Alder: Well as long as it still has some life, I'll look into it.
[14:42] Rob Linden: sounds good
[14:42] Rob Linden: anything else?
[14:42] Mm Alder: Oh, how about the compiler for Linden meshes?
[14:42] Soft Linden: Wow, I haven't seen any new details on that.
[14:42] Aimee Trescothick: the transcript for the session earlier will probably pop up at http://www.orange-island.com/?p=1051 at some point MM
[14:43] Soft Linden: I'm pretty sure I put a watch on the issue for finding/distributing that
[14:43] Michelle2 Zenovka: i wanted to mention SVC-3146 as we have been working around this with viewer source code mods (the old openjpeg top corner jobs) and textures that need kicking in the texture fether to downlaod in a reasonable time. but it was grabbed an hour ago by Simon Linden
[14:44] Rob Linden: looks
[14:44] Aimee Trescothick: oops, wrong page, http://www.orange-island.com/?p=895
[14:45] Mm Alder: Thanks Aimee. I never heard about Orange Island. Looks interesting.
[14:45] Aimee Trescothick: :)
[14:45] Soft Linden: Michelle2 - looks like he's got two theories on that already, and is investigating.
[14:46] Michelle2 Zenovka: my money is on similar assumtions (identical code?) to the viewer was used in the server for the discard cut off calcuations ;-)
[14:47] Michelle2 Zenovka: i also guess that the tcp_rates for openjpeg and kdu are different and that the maps generated by the server using openjpeg use the same tcp_rates as we see in the viewer code
[14:47] Michelle2 Zenovka: hence a miss-match with the server rates and truncated discard levels etc
[14:48] Michelle2 Zenovka: i believe the server used to use kdu for the maps and changed, about the time this trouble was kicking off (but also when open jpeg was stablising for the viewer)
[14:49] Rob Linden: well, looks like it's in good hands.
[14:49] Michelle2 Zenovka: cool
[14:49] Michelle2 Zenovka: it would save some nasty viewer patches
[14:50] Rob Linden: let's see where we're at next week with that one. with enough advance warning on this topic, we might convince Simon to join us for one of these
[14:50] Michelle2 Zenovka: ok nice
[14:51] Rob Linden: anything else we should cover?
[14:51] Rob Linden: (pretty good meeting for not having an agenda)
[14:51] Rob Linden: (or at least, a very long one)
[14:51] Rob Linden: going once....
[14:51] Rob Linden: going twice.......
[14:52] Squirrel Wood: Tacos!
[14:52] Rob Linden: ok...thanks everyone!
[14:52] Michelle2 Zenovka: Thanks
[14:52] Mm Alder: Thanks Rob
[14:52] Soft Linden: Tacos are not a valid open source discussion item.
[14:52] Michelle2 Zenovka: GPL'd Tacos
[14:52] Soft Linden: Try an open-faced sandwich next time.
[14:52] Squirrel Wood: open sauce :p
[14:52] Michelle2 Zenovka: Tacos with open sauce?
[14:52] Squirrel Wood: ^^
[14:52] Soft Linden: Allllllrighty
[14:53] Soft Linden: Take care :>
[14:53] Squirrel Wood: more meetings, more work? ^^
[14:53] Michelle2 Zenovka: Bye everyone
[14:53] Aimee Trescothick: bye :)
[14:53] Khyota Wulluf: bie everyoen

Open Source Meeting/2008-10-02

Navigation menu

Search