#bitfighter IRC Log

Index Search ←Prev date Next date→

IRC Log for 2015-02-25

Timestamps are in GMT/BST.

00:06:38Watusimoto_ has joined
00:14:23Watusimoto_ Quit (Ping timeout: 240 seconds)
00:27:08Watusimoto_ has joined
00:29:06Watusimoto_hi
00:29:13raptorjo
00:29:17raptorerr... hi
00:30:30Watusimoto_fingers in the home position, please
00:31:04Watusimoto_so thanks for fixing the master -- now the motd changes on clients when updated. I could even test it!
00:31:18raptoroh good... and it was back-portable to 019 ok?
00:31:55Watusimoto_it was master-only
00:32:01Watusimoto_so back-porting was automatic
00:32:14Watusimoto_just curious; why do you prefer ++it to it++?
00:32:17raptorwow, an issue was actually resolved!
00:32:50raptoroh... honestly, that was what all the examples just did... also, i think it saves a cpu cycle or two
00:32:58raptorbut really, i like it++ for most things
00:33:39raptoralso some of the examples on deleting-while-iterating actually depended on one or the other form, so I just left it because i didn't want to do much thinking :)
00:33:46Watusimoto_I'm willing to bet almost anytning that x++ and ++x compile to exactly the same thing
00:33:56raptorthey don't - i've seen it different
00:34:05raptorit++ uses a temporary, i think
00:34:13Watusimoto_can't be!
00:34:39Watusimoto_let's find out: http://stackoverflow.com/questions/1812990/incrementing-in-c-when-to-use-x-or-x
00:34:40raptori bet advanced compilers switch the it++ to ++it
00:35:06Watusimoto_the only cases I've seen ++x be useful is when the incrementing is abused; like this:
00:35:11Watusimoto_y = (++x)
00:35:22Watusimoto_which would give a different result from y = (x++)
00:35:25raptorha
00:35:35Watusimoto_but neither of which should _ever_ be used, IMO
00:35:35raptori've seen almost always used more in older C code
00:35:40raptorthe ++it
00:36:35Watusimoto_and please note that in a for loop, on primatives, there is absolutely no difference. Many coding styles will recommend never using an increment operator where it could be misunderstood; i.e., x++ or ++x should only exist on its own line, never as y=x++.
00:36:49Watusimoto_^^ totally agree with that
00:36:59raptoryes, i agree, too
00:37:06raptori hate having to do a double take on the code
00:37:07Watusimoto_And if used on its own line, the generated code is almost sure to be the same.
00:38:29raptorhere's one answer: https://stackoverflow.com/a/24904
00:39:35Watusimoto_hmmm.... >>> you got the iterator stuff exactly the wrong way round. ++i is faster, because it will not copy the object, i++ is slower, because it first creates a copy of the object, increments the iterator and then return the copied object (which is still in the same state)
00:40:11raptorso in class-operator scenarios it's slower, but i bet most c compilers on primitive types are equally efficient
00:41:01Watusimoto_I never use iterators, (at least no explicitly), so my thinking on them may not be correct
00:41:41raptorjava 1.4 code had iterators everywhere
00:42:09Watusimoto_I've never felt comfortable with them
00:42:10raptornow java uses this syntax for loops, which i like much better: for(String s: someStringList)
00:42:13Watusimoto_give me a for loop any day!
00:42:20Watusimoto_I like it better too
00:42:40Watusimoto_c# lets you do for(var s in someStringList)
00:42:50Watusimoto_i.e. you can let the compiler figure out what type s is
00:42:57raptoryep, much nicer
00:42:59raptoroh, ha!
00:43:11raptorc# is pseudo-dynamically typed?
00:43:22Watusimoto_handy when you are dealing with complex<nested<generic,types>>
00:43:29raptorthat's like python...
00:43:39raptorhaha
00:43:43Watusimoto_it's completely statically typed
00:43:54Watusimoto_var only works when the compiler can figure out the type from the statement
00:43:59raptorah ok
00:44:03Watusimoto_it reduces pedantry
00:44:07raptoryeah
00:44:17Watusimoto_no more: List<int> x = new List<int>()
00:44:24Watusimoto_rather, var x = new List<int>()
00:44:32raptorjava, at least up to java 1.7, always throws warnings when you don't specify the types
00:44:48Watusimoto_but you cannot do: var x;
00:44:52raptordrives me nuts when porting older code
00:44:58Watusimoto_yes
00:45:07Watusimoto_if the compiler can do the work, it should do the work
00:45:09raptortrue
00:45:20raptori think c# has surpassed java in many ways
00:45:25Watusimoto_yes
00:45:47Watusimoto_and it's (now) about the same level of open-sourcedness
00:45:49raptorand now that they've open sourced the VM... maybe we'll be a c# shop in the near future!
00:45:53Watusimoto_yes
00:45:54raptoryes
00:45:57Watusimoto_yes
00:45:59raptorhaha
00:46:14Watusimoto_neither seems truly opensource
00:46:19Watusimoto_only somewhat so
00:47:21raptoryeah, it's true - java spec has been open source for a while, but not the oracle/sun JVM
00:47:30raptorwhich is the fastest
00:47:55raptororacle has done a good job at slowly killing java's potential, i think
00:48:15raptorand google has done a good job at building it (with android)
00:48:24raptorback in a bit...
01:33:30raptornot sure if you've seen this: https://www.youtube.com/watch?v=8F9jXYOH2c0
01:39:57Watusimoto_man, this latest checkin was a waste of time!
01:40:02BFLogBot_ Commit: b2444475d5 | Author: watusimoto | Message: Whitespace
01:40:04BFLogBot_ Commit: f1a226eb50 | Author: watusimoto | Message: False start
01:40:05BFLogBot_ Commit: 7ca9641a9f | Author: watusimoto | Message: Variable name
01:40:07BFLogBot_ Commit: f330cb1513 | Author: watusimoto | Message: Switch motd from const char to string
01:40:08BFLogBot_ Commit: f2abccec84 | Author: watusimoto | Message: Make changing MOTD appear smoother on client (020 only, could be backported)
01:40:10BFLogBot_ Commit: 2dc697c013 | Author: watusimoto | Message: Merge
01:40:18Watusimoto_but it satisifies my inner Felix Unger (and now you know the reference!)
01:40:23raptorhaha
01:40:27raptori'll take a look..
01:41:11Watusimoto_You'll have to compile 020 with a modified version.h to fake it being 019d -- otherwise you don't get the current motd, and that's the only one that dynamically updates
01:41:29raptorwas the smoothness really in question?
01:42:24Watusimoto_not really
01:42:41Watusimoto_in 019d, the position resets when a new motd arrives
01:42:56raptorahhh
01:42:59Watusimoto_in practice, this is a very rare event, but it shows up a lot when testing it :-)
01:43:05raptorha!
01:43:21Watusimoto_now it just swaps out the old, and the overall position doesn't change
01:43:29Watusimoto_so it is pretty subtle
01:43:56raptoroh, if you didn't see... your Felix Unger commit had some muphrey's law aspect to it (take a close look): https://code.google.com/p/bitfighter/source/detail?r=ea54f970b3b0cdb4ab90ad3bb1f3c996d81fde2a
01:44:05raptor:)
01:47:40Watusimoto_great soccer
01:48:53Watusimoto_oh, great.
01:50:01BFLogBot_ Commit: 5343c377fb | Author: watusimoto | Message: @#$%^&
01:50:16Watusimoto_You probably know what that is without even looking :-)
01:50:31raptorhaha
01:59:56raptorI have 90% of an Opus implementation that I started before thanksgiving... trying to get the willpower to finish it
02:16:18raptorsleepy... good night!
02:16:23Watusimoto_night!
02:16:28raptor Quit ()
03:57:19BFLogBot_ Commit: 02e6cf105d | Author: watusimoto | Message: Replace color with name
03:57:21BFLogBot_ Commit: e8d082616a | Author: watusimoto | Message: Share code more
05:11:04Watusimoto_ Quit (Ping timeout: 245 seconds)
08:13:13Darrel Quit (Read error: Connection reset by peer)
08:13:55Darrel has joined
08:20:27Darrel Quit (Ping timeout: 256 seconds)
08:20:34Darrel has joined
08:31:06Darrel Quit (Ping timeout: 264 seconds)
08:45:13koda has joined
09:00:26Darrel has joined
09:52:17koda Quit (Quit: koda)
09:52:36koda has joined
10:32:49Watusimoto_ has joined
10:34:20CrazyLinuxNerd Quit (Quit: Going.. Going.. Gone)
10:39:23Watusimoto_ Quit (Ping timeout: 240 seconds)
10:44:57CrazyLinuxNerd has joined
11:16:39raptor has joined
11:16:39ChanServ sets mode +o
11:16:45raptorgood day!
11:40:10raptorHelpItemManager.cpp is broke with the GuageRenderer changes last night: it uses some previously public members of EnergyGaugeRenderer
11:40:22raptornot sure how to fix yet...
12:23:17koda Quit (Quit: koda)
12:23:42koda has joined
13:00:19watusimotohello
13:00:30watusimotoI'll fix helpItemRenderer
13:00:34watusimotosince I broke it
13:00:58watusimotoI moved on to a long-term project to get ready for porting to gles2
13:01:20watusimotoI did a bunch of mindless work, then wanted to look something up
13:01:23watusimotoand found this:
13:01:24watusimotohttps://github.com/p3/regal
13:01:32watusimotothis might be a good short-term solution
13:12:16raptorI've already loked at regal - it's overkill, unless we can pull out what we need
13:12:26raptoralso, kaen has rebased his shader work on 020 branch
13:13:53raptorthere's a diff of it somewhere; i'll get it after my next meeting
13:23:11kaen_mbp has joined
13:34:52watusimotomy biggest worry is not the shaders, but rather all the pushMatrix/popMatrix stuff; I started to rewrite that to do the management ourselves, but I was thinking regal might save us the trouble.
13:51:48kaen_mbp Quit (Ping timeout: 256 seconds)
14:01:20raptorthe problem I had with regal, when researching it, was that it added something like 8MB of library size
14:01:38raptorbut admittedly, i did not looking into specific build options to reduce that
14:10:32kaen_mbp has joined
14:20:25kaen_mbp_ has joined
14:22:01kaen_mbp Quit (Ping timeout: 250 seconds)
15:20:30kaen_mbp has joined
15:21:49kaen_mbp_ Quit (Ping timeout: 250 seconds)
15:28:03koda Quit (Ping timeout: 244 seconds)
15:36:08LordDVG has joined
15:50:00watusimotoI'll bet we could delete 90% of regal and still get what we need
16:41:55Nothing_Much Quit (Ping timeout: 264 seconds)
16:46:58CrazyLinuxNerd Quit (Quit: Going.. Going.. Gone)
16:47:16CrazyLinuxNerd has joined
17:07:22Nothing_Much has joined
17:28:29LordDVG Quit (Remote host closed the connection)
17:58:34koda has joined
18:06:57Darrel Quit (Read error: Connection reset by peer)
18:07:13Darrel has joined
19:26:00koda Quit (Read error: Connection reset by peer)
19:26:06kodabb has joined
20:30:49raptor Quit (Disconnected by services)
20:31:09raptor has joined
20:31:09ChanServ sets mode +o
21:39:37Nothing_Much Quit (Quit: Konversation terminated!)
22:12:10kodabb Quit (Quit: kodabb)
22:16:19watusimoto Quit (Quit: Leaving.)
23:01:40watusimoto has joined
23:01:40ChanServ sets mode +o
23:20:10raptorwe'd have to link against regal instead of the normal opengl libraries... at least, that's what it says it usage is
23:20:22raptorand we'd have to do nothing different in our code
23:22:27raptorlooks like oglconsole may not work with it..
23:28:55Nothing_Much has joined
23:30:59raptorregal's build system seems to compile in lots of dependencies, statically
23:34:45raptoroh wow, just the GL/Regal.h header is 2.5MB large
23:43:59raptorah ha - this may be more what we want instead of regal: https://github.com/lunixbochs/glshim
23:51:36Flynnn has joined
23:59:10raptori think glshim looks quite good - the openpandora guys developed it and have been running linux opengl games with it on their platform

Index Search ←Prev date Next date→

These logs were automatically created by BFLogBot on irc.freenode.net.