| 00:14.35 | *** join/#riftuidev Imhothar_ (miranda@ppp-188-174-14-226.dynamic.mnet-online.de) |
| 00:43.30 | *** join/#riftuidev Imhothar_ (miranda@ppp-188-174-156-127.dynamic.mnet-online.de) |
| 01:10.54 | *** join/#riftuidev Imhothar (miranda@ppp-188-174-71-176.dynamic.mnet-online.de) |
| 01:12.52 | *** join/#riftuidev ZorbaBeta (tocohl@maximilian.pavlovian.net) |
| 01:19.10 | *** join/#riftuidev Imhothar (miranda@ppp-188-174-159-85.dynamic.mnet-online.de) |
| 01:23.33 | *** join/#riftuidev Imhothar_ (miranda@ppp-188-174-117-246.dynamic.mnet-online.de) |
| 01:37.49 | *** join/#riftuidev Imhothar (miranda@ppp-188-174-118-226.dynamic.mnet-online.de) |
| 01:39.32 | *** join/#riftuidev seebs (~seebs@home.seebs.net) |
| 01:39.32 | *** mode/#riftuidev [+o seebs] by ChanServ |
| 01:42.04 | *** join/#riftuidev Imhothar_ (miranda@ppp-93-104-178-16.dynamic.mnet-online.de) |
| 01:43.42 | *** join/#riftuidev Imhothar_ (miranda@ppp-188-174-112-188.dynamic.mnet-online.de) |
| 02:29.57 | *** join/#riftuidev Torhal|work (~torhal_wo@76.164.170.2) |
| 02:29.57 | *** mode/#riftuidev [+v Torhal|work] by ChanServ |
| 03:05.08 | seebs | Agh, what a test case. |
| 03:05.13 | seebs | Got a test case for an alleged compiler bug. |
| 03:05.36 | seebs | Turns out that there exists a computation such that (int) 15 - EXPR is probably 5, but might be 4. |
| 03:05.49 | seebs | But if you store EXPR in a double, 15 - d is always exactly 5. |
| 03:06.52 | Ivnedar | which compiler? |
| 03:07.01 | seebs | gcc. |
| 03:07.09 | seebs | Compiler's right, test case is wrong. |
| 03:07.24 | seebs | The value computed is 200 * N, where N is very very close to 0.05. |
| 03:08.08 | seebs | Close enough that, when the rounding error is next to something with magnitude 8 or higher, it falls off the end of the representation. |
| 03:11.18 | Ivnedar | is the test case trying to find the precision bugs? |
| 03:13.56 | seebs | No, it's just extracted from some other test case. |
| 03:14.48 | seebs | And my general thought is, if you do floating point math that involves unrepresentable values, come out within .000001 of an integer value, and then convert the results to int by casting, you deserve whatever you get. |
| 03:15.08 | Ivnedar | pretty much :) |
| 04:28.24 | odine | how is new event stuff? |
| 04:28.53 | odine | i have nerve dmg in left hand... stuck w one hand atm lmao |
| 04:29.09 | odine | so slowwwwwwww responses |
| 04:29.25 | odine | but ne1 try em on pts yet? |
| 04:54.25 | seebs | Haven't tried yet. |
| 04:54.31 | seebs | Wow, nerve damage sucks. RSI? |
| 04:58.09 | odine | not sure yet |
| 04:58.21 | odine | have sling on |
| 04:58.29 | odine | on steroids |
| 04:58.58 | odine | if not better in...3 dats then neurologist i tink |
| 05:33.25 | *** part/#riftuidev odine (odine@cpe-74-77-173-83.buffalo.res.rr.com) |
| 05:37.32 | Ivnedar | ooh, nice keyfocus workaround ZorbaBeta |
| 06:59.37 | Baanano | :O 12 new openings on Redwood, has any other team moved there or is the RIFT one planning something big? |
| 07:16.25 | seebs | That's sort of odd given how recent the layoffs were. Unless finances changed or something. |
| 09:09.33 | *** join/#riftuidev Torhal|work (~torhal_wo@76.164.170.2) |
| 09:09.33 | *** mode/#riftuidev [+v Torhal|work] by ChanServ |
| 10:50.10 | *** join/#riftuidev Imhothar (miranda@ppp-188-174-143-230.dynamic.mnet-online.de) |
| 11:14.17 | *** join/#riftuidev Imhothar (miranda@ppp-188-174-143-230.dynamic.mnet-online.de) |
| 18:34.45 | *** join/#riftuidev odine (~odine@cpe-74-77-173-83.buffalo.res.rr.com) |
| 18:34.51 | odine | morning |
| 18:48.59 | odine | i need to get on pts and play with these events even ifi i opnly have one good hand lol |
| 19:01.22 | *** join/#riftuidev Imhothar (miranda@ppp-188-174-143-230.dynamic.mnet-online.de) |
| 19:40.31 | Imhothar | ZorbaTHut what kind of compiler are you using for the C++ stuff of Rift? |
| 19:42.55 | Imhothar | for release |
| 20:50.09 | bctrainers | Not sure if he is allowed to state that due to contract limitations :x |
| 21:16.34 | ZorbaBeta | I actually have no idea if I am or not, so I'm not gonna. Sorry :V |
| 21:17.42 | ZorbaBeta | I dunno if everyone's seen this yet or not, but http://forums.riftgame.com/rift-general-discussions/general-discussion/352460-tech-talk-rift-performance-2-2-a.html |
| 21:30.40 | seebs | ZorbaBeta: Wow. |
| 21:30.42 | seebs | Very cool. |
| 21:30.55 | seebs | I have a suggestion, which I think is probably of potential use to that project. |
| 21:31.00 | seebs | Well, a family of suggestions. |
| 21:31.16 | seebs | The high-expense, high-payoff one: Implement something like CoH's /demorecord. |
| 21:31.42 | seebs | Which allows you to record, not the generated pixels, but everything the client's rendering is going to do, for a period of time. And then (in theory) play back that log. |
| 21:32.28 | seebs | One of the things that would be stupendously useful for performance tweaking would be usable benchmarks -- things you could run on multiple players' computers, and compare results. This would be a way to get them. |
| 21:33.28 | seebs | Thing is, right now, there's no way to make sense of comparisons of reported frame rates, because no one's doing the same things, or seeing the same things. |
| 21:34.00 | seebs | I suppose an alternative would be the Chronicle Of Mind Control, in which the PC is run through a completely scripted event with no ability to interact or change behaviors, and some performance logging. |
| 21:57.23 | ZorbaBeta | seebs, I've actually had something like that on my wishlist for a while, but . . . let's just say it's not exactly simple |
| 21:57.38 | ZorbaBeta | it's one of those things that's really easy if the engine is designed with it in mind, and nigh-impossible if it's not. in our case, sadly, it's not. |
| 21:57.58 | seebs | Yeah. |
| 21:58.29 | ZorbaBeta | I have an idea that *might* end up being vaguely equivalent, which I might end up implementing as part of the performance work. but unless it ends up being a massive gain in and of itself, it's probably not gonna happen :) |
| 21:59.41 | seebs | I have just gotten a very harsh reminder of how vital it can be to actually stop and develop useful test cases and test assumptions about performance. |
| 22:03.50 | ZorbaBeta | oh? |
| 22:05.37 | seebs | My pseudo performance stuff. |
| 22:06.13 | seebs | We've spent most of a YEAR putting up with builds taking forever and completely crippling machines while running, and we never stopped to actually verify where the performance hit was. |
| 22:06.39 | seebs | It turned out to take maybe three days' work to produce a 20-50% improvement in our real use cases. |
| 22:09.53 | ZorbaBeta | one of my coworkers at google had a similar event, when he decided to rewrite how the build tools worked because he thought he could do better |
| 22:09.57 | ZorbaBeta | ended up being a massive improvement |
| 22:10.06 | ZorbaBeta | as a reward he ended up working on tools for a few years until he could escape it :V |
| 22:10.12 | seebs | heh |
| 22:10.16 | seebs | See, I *like* tools work. |
| 22:10.21 | seebs | It's what usually makes me happiest. |
| 22:10.37 | seebs | Next on my list, I think I can dramatically reduce the time for the initial project configuration stage, which currently takes far too long. |
| 22:13.40 | ZorbaBeta | I'll admit that a while back I also got fed up with build tools and ended up writing my own :V |
| 22:20.41 | seebs | Heh, I'm messing with an SDK for mobile stuff (pretty sure I am gonna rewrite the Miracle Modus in it), and I've already sent them a build system patch. |
| 22:21.03 | seebs | I observed to coworkers that I was really proud of myself. I used their dev tools for 6-10 hours before I sent my first suggested build system patch. |
| 22:24.49 | ZorbaBeta | someday I really need to polish up and document the build system I ended up with |
| 22:25.16 | ZorbaBeta | working on it was kinda the first time my api design sensibilities really started to gel. it had a surprising amount of influence on rift :V |
| 23:09.52 | ZorbaBeta | unrelated: I purchased a desk-mount monitor arm solely so the cat could sleep under my left monitor. it is working beautifully :3 |
| 23:13.54 | Imhothar | Oh man, I know wway more about ancient mythologies now than I ever anticipated |
| 23:15.02 | Imhothar | went through about 1000 deities for my project |
| 23:15.14 | Imhothar | it's insane what those people came up with |
| 23:19.06 | ZorbaBeta | that actually sounds pretty awesome, I've always liked mythology |
| 23:19.22 | Imhothar | yah, it's very interesting |
| 23:20.06 | Imhothar | it really refelcts the cultural differences. In norse mythology most deities were warriors or womehow related to war |
| 23:21.00 | Imhothar | oh, and the norse names are barely pronouncable |
| 23:21.36 | Imhothar | stuff like Skordrjakap, with rolled R |
| 23:21.51 | Imhothar | and yes, that's "rdr" they got those alot |
| 23:22.19 | Imhothar | many names end with -dr |
| 23:26.20 | Imhothar | Oh and they still speak that way in Iceland |
| 23:40.11 | *** join/#riftuidev Noshei (~quassel@c-24-9-171-99.hsd1.co.comcast.net) |
| 23:55.22 | Imhothar | Hmm the main Aztec god "invented himself and created himself" |
| 23:55.25 | Imhothar | wonder how that works |
| 23:58.12 | Noshei | "we seem to be using too much CPU, ironically, on one of our performance optimization passes." - ZorbaBeta |
| 23:58.26 | Noshei | lol, that one line was great |