[Back to Index]

  
[00:03] Littleboy (~littleboy@c-174-62-174-218.hsd1.ma.comcast.net) got lost in the net-split.
[00:07] <-- Vanfanel left irc: Quit: Lost terminal
[00:11] <-- D0SFreak left irc: Ping timeout: 246 seconds
[00:40] <-- criezy left irc: Quit: criezy
[00:56] --> RUBICN64 joined #scummvm.
[01:12] <-- Dominus left irc: Ping timeout: 246 seconds
[01:12] --> Dominus joined #scummvm.
[01:17] <-- Javacat left irc: Quit: Please, try the fish
[01:22] <-- _marc` left irc: Ping timeout: 260 seconds
[01:26] <-- kurtwr2 left irc: Ping timeout: 272 seconds
[01:37] --> _marc` joined #scummvm.
[01:38] <-- RUBICN64 left irc: Quit: AndroidIrc Disconnecting
[01:41] --> Vampire0_ joined #scummvm.
[01:45] <-- Guest72212 left irc: Ping timeout: 272 seconds
[02:22] --> kurtwr joined #scummvm.
[03:22] --> RUBICN64 joined #scummvm.
[03:31] --> ImperatorPrime joined #scummvm.
[03:34] --> LordHoto joined #scummvm.
[03:34] #scummvm: mode change '+o LordHoto' by ChanServ!ChanServ@services.
[03:43] <-- Ramal left irc: Ping timeout: 272 seconds
[03:44] --> Ramal joined #scummvm.
[03:44] <-- RUBICN64 left irc: Quit: AndroidIrc Disconnecting
[04:00] <-- MrSiggler left irc: Quit: Textual IRC Client: www.textualapp.com
[04:01] --> Poly-C joined #scummvm.
[04:04] <-- Polynomial-C left irc: Ping timeout: 245 seconds
[04:28] <-- LordHoto left irc: Quit: leaving
[04:49] <-- SugarCube left irc: Ping timeout: 272 seconds
[05:00] --> SugarCube joined #scummvm.
[05:02] --> bgKa joined #scummvm.
[05:02] #scummvm: mode change '+o bgKa' by ChanServ!ChanServ@services.
[05:25] <-- pigeon left irc: Ping timeout: 250 seconds
[05:49] --> uruk-hai joined #scummvm.
[05:49] #scummvm: mode change '+o uruk-hai' by ChanServ!ChanServ@services.
[05:57] <-- Cheeseness left irc: Quit: Leaving.
[06:03] --> waltervn joined #scummvm.
[06:04] <waltervn> morning
[06:04] <-- _marc` left irc: Ping timeout: 250 seconds
[06:44] <-- Lightkey left irc: Ping timeout: 260 seconds
[06:48] <-- bgKa left irc: Ping timeout: 272 seconds
[06:55] <-- _sev left irc: Quit: This computer has gone to sleep
[06:57] --> Lightkey joined #scummvm.
[07:02] --> _marc` joined #scummvm.
[07:03] <-- SylvainTV left irc: Ping timeout: 258 seconds
[07:13] --> pigeon joined #scummvm.
[07:13] #scummvm: mode change '+v pigeon' by ChanServ!ChanServ@services.
[07:33] --> ajax16384 joined #scummvm.
[07:33] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[08:03] --> Cheeseness joined #scummvm.
[08:29] <-- Jon_God left irc: Remote host closed the connection
[08:41] <fuzzie> gsoc tshirt this year is better than I expected, neat
[08:48] <waltervn> how did gsoc turn out this year?
[08:56] <uruk-hai> fuzzie, yes :D
[08:56] <uruk-hai> waltervn, pretty good ;)
[08:56] <uruk-hai> i think we were as successfull as last year, if not more
[08:57] <uruk-hai> at least now i was able to finish CGE2... :D
[08:59] <waltervn> Good. :) Though, to me personally, any GSoC that doesn't include the "Audio Output Selection" and "MIDI Device Configuration" tasks, is a big disappointment. ;)
[09:01] <waltervn> we should just put next cool stuff on the list next year, then the student will be forced to pick those
[09:01] <waltervn> *fewer cool stuff
[09:01] <waltervn> *less
[09:01] --> _sev|work_ joined #scummvm.
[09:01] #scummvm: mode change '+o _sev|work_' by ChanServ!ChanServ@services.
[09:03] <-- _sev|work left irc: Ping timeout: 245 seconds
[09:04] <uruk-hai> :D
[09:04] <uruk-hai> speaking about it, i have a friend at university who is really interested in things like these
[09:04] <uruk-hai> i won't forget to direct him to our ideas page next semster ;)
[09:08] --> TMM joined #scummvm.
[09:08] --> Strangerke|work joined #scummvm.
[09:08] #scummvm: mode change '+o Strangerke|work' by ChanServ!ChanServ@services.
[09:08] <Strangerke|work> hi guys
[09:08] <waltervn> hey
[09:09] <uruk-hai> heya :)
[09:36] <uruk-hai> guys! can someone help me a bit with our advanced detector? here's my problem: both soltys and sfinx has vol.* files what are checked when detecting for these games. the problem is that when i "add game.." sfinx (however, thsi problem doesn't manifest in soltys), it brings up a list to choose from sfinx or soltys. if i choose sfinx, it all works perectly, i just want to avoid this list bringing up
[09:37] <uruk-hai> so i decided to check for a certain file in sfinx's data archives, vol.dat
[09:37] <uruk-hai> a file that is not amongst soltys data for sure
[09:37] <uruk-hai> but i don't know where i can do this check
[09:38] <uruk-hai> i suppose i'll have to override one of the functions of AdvancedMetaEngine, i just don't know which one is it
[09:39] <Strangerke|work> iirc, there's some pretty insane detection code in Gob, you could give it a look
[09:40] <DrMcCoy> There's just a weird detector for Once Upon a Time
[09:41] <DrMcCoy> https://github.com/scummvm/scummvm/blob/master/engines/gob/detection/detection.cpp#L84
[09:43] <uruk-hai> ah, okay, tahnks guys :)
[09:59] <-- Cheeseness left irc: Quit: Leaving.
[10:05] Action: uruk-hai summons Strangerke|work
[10:24] <-- _sev|work_ left irc: Quit: This computer has gone to sleep
[10:25] --> _sev|work_ joined #scummvm.
[10:25] <-- _sev|work_ left irc: Changing host
[10:25] --> _sev|work_ joined #scummvm.
[10:25] #scummvm: mode change '+o _sev|work_' by ChanServ!ChanServ@services.
[10:27] --> D0SFreak joined #scummvm.
[10:28] --> bgKa joined #scummvm.
[10:28] #scummvm: mode change '+o bgKa' by ChanServ!ChanServ@services.
[10:33] --> Deledrius_ joined #scummvm.
[10:33] --> h00ligan joined #scummvm.
[10:33] #scummvm: mode change '+o h00ligan' by ChanServ!ChanServ@services.
[10:36] <-- ajax16384 left irc: Ping timeout: 258 seconds
[10:36] <-- Deledrius left irc: Ping timeout: 258 seconds
[10:45] --> KingGuppy joined #scummvm.
[10:46] <-- GodGuppy left irc: Quit: No Ping reply in 180 seconds.
[10:55] --> Raziel^ joined #scummvm.
[10:55] #scummvm: mode change '+v Raziel^' by ChanServ!ChanServ@services.
[11:11] --> SylvainTV joined #scummvm.
[11:11] #scummvm: mode change '+o SylvainTV' by ChanServ!ChanServ@services.
[11:13] --> kaan joined #scummvm.
[11:31] <-- echelog-1 left irc: Ping timeout: 246 seconds
[11:33] <-- bgKa left irc: Ping timeout: 250 seconds
[11:35] --> echelog-1 joined #scummvm.
[11:37] <-- ImperatorPrime left irc: Quit: Konversation terminated!
[11:50] <wjp> about that putt-putt gm/mt32 bug report: is there an "obvious" thing that would make the user conclude that "ScummVM considers they are MT32"?
[11:51] <wjp> because I'm not sure which symptoms would make you readily conclude that
[11:52] <Raziel^> hey wjp
[11:54] <wjp> hi
[11:55] <wjp> FWIW, no objections for a new drawer icon of course
[11:55] <Raziel^> thank you :-)
[11:55] <wjp> although having a top-level icons/ directory is starting to look a bit strange
[11:55] <wjp> (but that has no immediate impact on this PR)
[11:58] <Raziel^> i'm writing an installer script too for the next release, still hvae to figure out how to tell it to install that drawer a level above the other files :-)
[12:02] --> GitHub192 joined #scummvm.
[12:02] <GitHub192> [scummvm] urukgit pushed 1 new commit to master: http://git.io/4_lofw

[12:02] GitHub192 (GitHub192@192.30.252.45) left #scummvm.
[12:02] <wjp> uruk-hai: such commits could really use an explanation of why in the commit message
[12:03] <uruk-hai> oh, sorry :\
[12:04] <uruk-hai> btw it's to avoid clashing with CGE2's detection and also it's really not needed anymore
[12:19] <-- uruk-hai left irc: Quit: Leaving
[12:20] --> GitHub74 joined #scummvm.
[12:20] <GitHub74> [scummvm] digitall closed pull request #501: AMIGAOS: Add and make use of a new drawer icon (master...patch-3) http://git.io/ahYV9g
[12:20] GitHub74 (GitHub74@192.30.252.40) left #scummvm.
[12:20] --> GitHub93 joined #scummvm.
[12:20] <GitHub93> [scummvm] digitall pushed 2 new commits to master: http://git.io/61weaQ
[12:20] <GitHub93> scummvm/master 01dc551 raziel-: AMIGAOS: Make use of the new drawer icon...
[12:20] <GitHub93> scummvm/master 2b9a5ca David Turner: Merge pull request #501 from raziel-/patch-3...
[12:20] GitHub93 (GitHub93@192.30.252.43) left #scummvm.
[12:20] <Raziel^> thank you :-)
[12:20] <Raziel^> can i delete the branch?
[12:21] <Raziel^> and the two others aswell?
[12:21] Nick change: _sev|work_ -> _sev|work
[12:23] <droid2727> Raziel^: "two others"?
[12:24] <Raziel^> droid2727: sorry, i meant the other two branches (patch-1 and patch-2) from former PR's
[12:24] <droid2727> yes, you can delete them
[12:24] <Raziel^> ok, thank you
[12:26] <-- kaan left irc: Ping timeout: 260 seconds
[12:29] <droid2727> wjp: he had "don't use general midi" selected?
[12:56] <-- droid2727 left irc: Ping timeout: 272 seconds
[13:06] <-- Kirben left irc:
[13:08] --> droid2727 joined #scummvm.
[13:08] #scummvm: mode change '+o droid2727' by ChanServ!ChanServ@services.
[13:10] --> kaan joined #scummvm.
[13:20] --> meunierd_ joined #scummvm.
[13:34] --> MrSiggler joined #scummvm.
[14:23] <-- Strangerke|work left irc: Quit: Bbl
[14:27] --> dtcrshr joined #scummvm.
[14:36] --> RichieSams|work joined #scummvm.
[14:36] #scummvm: mode change '+o RichieSams|work' by ChanServ!ChanServ@services.
[14:44] <-- _marc` left irc: Ping timeout: 272 seconds
[15:45] --> _marc` joined #scummvm.
[15:48] <DrMcCoy> Ooooh
[15:48] <DrMcCoy> https://github.com/keedreams/keen
[15:49] <DrMcCoy> "This repository contains the source for Commander Keen in Keen Dreams. It is released under the GNU GPLv2+."
[15:49] <-- Raziel^ left irc: Read error: Connection reset by peer
[15:49] --> Raziel^ joined #scummvm.
[15:49] #scummvm: mode change '+v Raziel^' by ChanServ!ChanServ@services.
[15:55] --> RUBICN64 joined #scummvm.
[16:05] <-- RUBICN64 left irc: Quit: AndroidIrc Disconnecting
[16:08] Nick change: Vampire0_ -> Vampire0
[16:10] <-- D0SFreak left irc: Ping timeout: 260 seconds
[16:14] <Strangerke> nice :D
[16:21] --> WooShell joined #scummvm.
[16:22] <WooShell> meow =^.^=
[16:35] --> Littleboy joined #scummvm.
[16:35] #scummvm: mode change '+o Littleboy' by ChanServ!ChanServ@services.
[16:38] <RichieSams|work> can someone help me understand which compilers support 64 bit byte swapping?
[16:38] <RichieSams|work> vs, which ones need a macro definition
[16:41] --> bgKa joined #scummvm.
[16:41] #scummvm: mode change '+o bgKa' by ChanServ!ChanServ@services.
[16:45] --> D0SFreak joined #scummvm.
[16:53] --> Wishmaster joined #scummvm.
[16:56] <somaen> RichieSams|work: Why do you need that information?
[16:57] <RichieSams|work> to add to endian.h
[16:57] <RichieSams|work> to support 64bit swaps
[16:58] <RichieSams|work> in order to support stream out of int64 and after some additions to the stream class, float and double
[16:59] <droid2727> I really dislike adding the 64-bit functions
[16:59] <droid2727> both the macros and the Stream functions
[17:00] <droid2727> 64-bit integers aren't supported on every platform we support
[17:01] <droid2727> (iirc)
[17:02] <RichieSams|work> _sev checked
[17:02] <RichieSams|work> and thought they were
[17:02] <RichieSams|work> where would I check to find out?
[17:05] <droid2727> By checking every toolchain
[17:06] <droid2727> But if uint64 is always defined now by configure, then we do assert 64-bit is available already
[17:07] --> Vanfanel joined #scummvm.
[17:08] <droid2727> (and then can ignore what I said altogether)
[17:12] <droid2727> 64-bit looks optional in configure still
[17:13] <Vanfanel> Is there someone who knows about the graphics backends? I usually ask LordHoto, but he's not here now
[17:13] <droid2727> If we want to make an assertion that 64-bit integers are available, then that should be changed first
[17:14] Action: Raziel^ summons LordHoto
[17:15] <fuzzie> Vanfanel: you should just ask and see if anyone responds, honestly
[17:16] <RichieSams|work> droid2727: ok
[17:16] <RichieSams|work> thanks for the info
[17:17] <RichieSams|work> I could disable the 64 bit sections with a #ifdef
[17:17] <RichieSams|work> using the flag from configure
[17:17] <Vanfanel> fuzzie: ok, sorry about that. That's nettiquette I guess.
[17:18] <droid2727> RichieSams|work: but then your engine would only work with that feature
[17:18] <RichieSams|work> true
[17:18] <droid2727> RichieSams|work: Which is why, IMO, you should avoid them
[17:19] <RichieSams|work> but would the backends the don't support 64 be able to play the engine at all?
[17:19] <RichieSams|work> since it requires doubles
[17:20] <RichieSams|work> I mean I guess you could have a FPU that supports doubles
[17:20] <RichieSams|work> and not have int64 support
[17:21] <Vanfanel> Ok, I have setup EGL context for GLES1 and GLES2 on the Raspberry Pi, and both are WAY slower (15% CPU usage vs 25%) than dispmanx (native Rpi 2D API) rendering, which I currently have working though my own SDL 1.2.x "hack". So I'm considering another solution, like a modified surfacesdl backend, etc. What other alternatives could I have?
[17:21] <DrMcCoy> RichieSams|work: Usually, toolchains for systems without hardware floating point support come with software implementations
[17:21] <DrMcCoy> Which are, of course, dead slow
[17:21] <-- TMM left irc: Quit: Ex-Chat
[17:22] <Vanfanel> What I would like is that Rpi users don't have to install an special libSDL implementation
[17:22] <RichieSams|work> but even software implementations of double would require int64, no?
[17:22] <-- kurtwr left irc: Ping timeout: 245 seconds
[17:22] <DrMcCoy> No
[17:23] <DrMcCoy> It's not even certain that a double is actually 64bit wide
[17:23] <Vanfanel> these percents are both in Monkey Island intro
[17:23] <RichieSams|work> would they massively truncate or do some kind of fixed point with two int32?
[17:24] <RichieSams|work> ah true, 64 bit wide is by IEEE standard
[17:24] <DrMcCoy> The standard only says that sizeof(double) is not smaller than sizeof(float)
[17:24] <RichieSams|work> oh!
[17:24] <DrMcCoy> sizeof(double) == sizeof(float) is entirely valid
[17:25] <RichieSams|work> /cry
[17:25] <fuzzie> Vanfanel: can't you just use dispmanx directly?
[17:26] <RichieSams|work> so I guess the hacky way would be, for 64 disabled backends, truncate double to floats
[17:27] <droid2727> how does that work for saves?
[17:27] <droid2727> isn't that what this is about?
[17:28] <RichieSams|work> yes
[17:28] <droid2727> still can't load a 64-bit integer if it's the type isn't present
[17:28] --> sirlemonhead joined #scummvm.
[17:28] <droid2727> -it's
[17:28] <DrMcCoy> RichieSams|work: Also, just dumping the float data to file is bad too
[17:28] <RichieSams|work> right
[17:28] <DrMcCoy> A float doesn't have to be IEEE
[17:29] <RichieSams|work> the plan is to manually serialize as IEEE
[17:29] <RichieSams|work> not cast
[17:29] <RichieSams|work> then manually deserialize
[17:30] <droid2727> and as I said days ago, do we want to assert ieee format across all platforms?
[17:30] <RichieSams|work> we don't
[17:30] <RichieSams|work> you would do it in software
[17:30] <RichieSams|work> not a cast
[17:31] <droid2727> you just said to serialize as ieee?

[17:31] <RichieSams|work> so, calculate the exponent and significand from the number, store as such (as an int)
[17:31] <RichieSams|work> then read as an int
[17:32] <RichieSams|work> bitshift to give you the significand and exponent
[17:32] <RichieSams|work> and manually calculate the float again
[17:33] --> Wishmaster joined #scummvm.
[17:34] <RichieSams|work> would that work?
[17:34] <RichieSams|work> granted, you would be adding a couple math ops per read
[17:34] <RichieSams|work> but how often do we read and write floating point data?
[17:36] <RichieSams|work> is the sizeof(float) also not set?
[17:36] <RichieSams|work> just sizeof(double) >= sizeof(float)
[17:38] <-- Raziel^ left irc: Ping timeout: 245 seconds
[17:38] <DrMcCoy> RichieSams|work: As far as I know only that
[17:39] <RichieSams|work> ah....
[17:39] <RichieSams|work> any thoughts on the manual serialization?

[17:40] --> Wishmaster joined #scummvm.
[17:43] <DrMcCoy> RichieSams|work: Possible. You can use frexp() to split the float, ldexp() to merge it again
[17:44] <RichieSams|work> yea, wjp suggested that the other day. I put together a quick demo and it seemed to work well
[17:47] <Vanfanel> fuzzie: use dispmanx directly in scummvm is what I would like to do now, but there are some difficulties involved. For example, I would still need to call SDL_SetVideoMode() so SDL initialices mouse input. Ok, I can call SDL_SetVideoMode() with a surface size of 0x0, but then, it still creates a visible surface with the monitor's native resolution. No idea why: 0x0 should just initialize input in this case. Then I would have to "
[17:55] --> kurtwr joined #scummvm.
[17:57] <-- MrSiggler left irc: Quit: My iMac has gone to sleep. ZZZzzz&
[18:04] --> _sev joined #scummvm.
[18:04] <-- _sev left irc: Changing host
[18:04] --> _sev joined #scummvm.
[18:04] #scummvm: mode change '+o _sev' by ChanServ!ChanServ@services.
[18:07] --> kurtwr2 joined #scummvm.
[18:08] <-- kurtwr left irc: Ping timeout: 272 seconds
[18:31] --> kurtwr joined #scummvm.
[18:31] <-- kurtwr2 left irc: Ping timeout: 240 seconds
[18:45] <-- Vanfanel left irc: Ping timeout: 240 seconds
[18:51] --> criezy joined #scummvm.
[18:51] #scummvm: mode change '+o criezy' by ChanServ!ChanServ@services.
[18:52] --> frankyboy_ joined #scummvm.
[19:03] --> GitHub150 joined #scummvm.
[19:03] <GitHub150> [scummvm] RichieSams opened pull request #508: COMMON: Add support for endian-safe reading/writing of int64 (master...add_endian_and_stream_support_for_int64) http://git.io/N3ftNw
[19:03] GitHub150 (GitHub150@192.30.252.34) left #scummvm.
[19:03] <RichieSams|work> droid2727: pull request created
[19:04] <RichieSams|work> I noted your thoughts, but feel free to add others

[19:16] --> Wishmaster joined #scummvm.
[19:17] <-- bgKa left irc: Ping timeout: 272 seconds
[19:20] <droid2727> RichieSams|work: But, can you do your floating point value storing without 64-bit?
[19:20] <droid2727> that's kind of the point I was trying to make
[19:20] <droid2727> avoid the need for a 64-bit requirement altogether
[19:21] <RichieSams|work> I don't see how. Other than with a string
[19:21] <RichieSams|work> unless there is a way to split it into two 32 bits
[19:25] <droid2727> That is why I suggested string last week
[19:25] <droid2727> even if it is "crazy"
[19:26] <droid2727> and as DrMcCoy said, there's no guarantee 64-bit will be enough
[19:26] <RichieSams|work> I guess if we let the string output have a decent amount of digits of precision, the truncation shouldn't be terrible
[19:30] <-- D0SFreak left irc: Ping timeout: 240 seconds
[19:43] --> uruk-hai joined #scummvm.
[19:43] #scummvm: mode change '+o uruk-hai' by ChanServ!ChanServ@services.
[19:49] <-- frankyboy_ left irc: Remote host closed the connection
[19:50] --> RUBICN64 joined #scummvm.
[19:50] <somaen> Is this for saving floats again?
[19:55] --> Sylvain joined #scummvm.
[19:57] <-- SylvainTV left irc: Ping timeout: 246 seconds
[20:00] <droid2727> yes
[20:07] <-- blitter left irc: Read error: Connection timed out
[20:08] --> blitter joined #scummvm.
[20:14] <Strangerke> SylvainTV: Stop hiding! I see you!
[20:26] <wjp> should be doable to store the significant in two steps as two uint32s
[20:27] <wjp> just shift, floor, cast to uint32, subtract, shift, and cast the rest
[20:28] <-- RUBICN64 left irc: Quit: AndroidIrc Disconnecting
[20:28] <RichieSams|work> ok, I've got working code that manually converts a double to a uint64
[20:29] <RichieSams|work> I could then shift and split this into two uint32
[20:29] <-- _sev left irc: Quit: This computer has gone to sleep
[20:29] <RichieSams|work> which would remove the need for the int64 stream and endian stuff
[20:31] <wjp> the tricky bit may be getting the infinity, NaN and sign subtleties right
[20:31] <RichieSams|work> the sign I've gotten
[20:31] <RichieSams|work> infinity and NaN I don't
[20:32] <RichieSams|work> http://pastebin.com/wJ1gYZig
[20:41] Nick change: Sylvain -> SylvainTV
[20:41] #scummvm: mode change '+o SylvainTV' by ChanServ!ChanServ@services.
[20:41] --> _sev joined #scummvm.
[20:41] <-- _sev left irc: Changing host
[20:41] --> _sev joined #scummvm.
[20:41] #scummvm: mode change '+o _sev' by ChanServ!ChanServ@services.
[20:43] <-- ny00123 left irc: Quit: Leaving
[20:44] --> ny00123 joined #scummvm.
[20:45] <-- h00ligan left irc: Read error: Connection reset by peer
[20:46] <-- uruk-hai left irc: Quit: Leaving
[20:46] --> D0SFreak joined #scummvm.

[20:51] --> Wishmaster joined #scummvm.
[20:52] <-- Wishmaster left irc: Client Quit
[20:54] <RichieSams|work> working on a split int32 solution
[21:00] --> Wishmaster joined #scummvm.
[21:03] <wjp> (you should do a s/significant/significand/ )
[21:10] <-- Wishmaster left irc: Read error: Connection reset by peer
[21:30] <-- pigeon left irc: Read error: Connection reset by peer
[21:31] --> pigeon joined #scummvm.
[21:31] #scummvm: mode change '+v pigeon' by ChanServ!ChanServ@services.
[21:42] <RichieSams|work> hmm
[21:42] <RichieSams|work> wjp: are you still here?
[21:43] <RichieSams|work> if I floor and subtract, the conversion becomes lossy
[21:43] <RichieSams|work> because of the subtraction
[21:45] <RichieSams|work> the other option is jujst to shift twice, without subtracting
[21:45] <RichieSams|work> however, that means you're relying on the cast from double to int to truncate properly
[21:45] <RichieSams|work> and I don't know if we can trust that
[21:45] <-- WooShell left irc: Quit: Walking upside down in the sky, between the satellites passing by. Gliding along the black rainbow, I fly away with my shadow. Scratching the moon like a DJ, the night follows its odyssey.
[22:00] <-- droid2727 left irc: Ping timeout: 260 seconds
[22:00] <wjp> RichieSams|work: I'm not sure I see the lossiness due to subtraction
[22:01] <wjp> but you can always just use the floor for the cast and keep the original value
[22:01] <wjp> uh, wait, I'm not sure I see that last one actually
[22:02] <DrMcCoy> dammit clones
[22:04] <RichieSams|work> say your original value is 12345.6789
[22:04] <RichieSams|work> you subtract to get 0.6789
[22:05] <RichieSams|work> the representation of 0.6789 is going to shift the precision
[22:06] <RichieSams|work> right, if you keep the original value and keep going
[22:06] <RichieSams|work> it will work
[22:06] --> droid2727 joined #scummvm.
[22:06] #scummvm: mode change '+o droid2727' by ChanServ!ChanServ@services.
[22:07] <RichieSams|work> however, now you're doing a uint32 secondValue = (uint32)veryLargeDouble
[22:07] <RichieSams|work> and relying on a truncate to get the 32 lower bits
[22:07] <-- pigeon left irc: Ping timeout: 240 seconds
[22:09] <RichieSams|work> http://pastebin.com/kewLhPKQ
[22:10] <RichieSams|work> wjp: The struct can be packed, I just left it big for simplicity atm
[22:11] <-- kaan left irc: Ping timeout: 260 seconds
[22:12] --> kaan joined #scummvm.
[22:16] <RichieSams|work> Large numbers are lossless
[22:16] <RichieSams|work> numbers smaller than 1.0 are lossy
[22:17] <RichieSams|work> though, the losses are really really small
[22:20] <-- _marc` left irc: Ping timeout: 272 seconds
[22:28] <wjp> I don't understand the purpose of the frexp calls in decodeDouble()
[22:30] <wjp> you're more or less arbitrarily shifting significandTwo now depending on the highest set bit
[22:31] <wjp> (also still note the spelling of significand)
[22:31] <wjp> but bedtime; good night
[22:31] <surix> funny that i was just discussing something similar to this. using integers to perform math for subpixel movement
[22:35] <surix> basically using values shifted left 16 bits. so 5 would be 0x50000 and .5 something like 0x08000 then to do 5 times .5 50000*8000 and shift right 16 to get 28000 (2.5)
[22:36] --> pigeon joined #scummvm.
[22:36] #scummvm: mode change '+v pigeon' by ChanServ!ChanServ@services.
[22:46] --> Javacat joined #scummvm.
[22:56] <-- criezy left irc: Quit: criezy
[22:57] <-- waltervn left irc: Quit: night!
[22:58] <RichieSams|work> wjp: the completed code
[22:59] <RichieSams|work> again, for very small numbers, the algorithm becomes lossy, but they're really small losses
[23:00] <-- RichieSams|work left irc: Quit: driving home
[23:02] sirlemonhead (bduncan22@95.44.164.183) left #scummvm.
[23:05] --> Kirben joined #scummvm.
[23:05] #scummvm: mode change '+o Kirben' by ChanServ!ChanServ@services.
[23:12] <-- _sev|work left irc: Read error: Connection reset by peer
[23:12] --> _sev|work joined #scummvm.
[23:12] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[23:14] --> Cheeseness joined #scummvm.
[23:30] droid2727 (droid2727@2600:1001:b00d:2b50:dd68:9c7d:f86:2844) left #scummvm ("Statement: Good-bye, meatbags!").
[23:42] --> Jon_God joined #scummvm.
[23:49] <-- D0SFreak left irc: Ping timeout: 255 seconds
[23:54] --> Vanfanel joined #scummvm.
[23:58] <-- meunierd_ left irc: Quit: meunierd_
[00:00] --- Wed Sep 17 2014