[Back to Index]
[00:32] <-- m_kiewitz left irc: Quit: technology isn't intrinsically good or evil. It's how it's used. Like the Death Ray.
[00:32] <-- SylvainTV left irc: Read error: Connection reset by peer
[00:50] <Deledrius> tsoliman, the wikipedia page does animate, as long as you're not using a Chrome-based browser because they've refused to implement it. >_<
[00:51] <Deledrius> Firefox works fine though.
[00:54] <snover> i try to avoid off-topic rants, but APNG is such a bunch of crap
[00:55] <snover> MNG was designed as the animated version of PNG, and was implemented in Firefox for a while, until one of the developers there complained that libmng was too big and deleted support for it
[00:55] <snover> then invented APNG instead
[00:56] <Lightkey> my homie
[00:58] <Lightkey> and after the libmng maintainer sized it down to a few dozen kBs, he gets told "it was never about the size", that was such a sad story all around
[00:58] <Lightkey> but I think we had that before here :-)
[01:00] <snover> 16 years later, still no widely available standard format for lossy compression with an alpha channel
[01:01] <snover> ok, rant over, thanks for listening :)
[01:01] <Lightkey> indeed, a point that was totally forgotten in the most voted bug on bugzilla, MNG also included JNG
[01:02] <snover> i found the bug, it took 2 minutes for bugzilla to send the page back&
[01:02] --> DominusExult joined #scummvm.
[01:06] <-- Dominus left irc: Ping timeout: 260 seconds
[01:07] Nick change: DominusExult -> Dominus
[01:15] <-- Mellified_Man_ left irc: Read error: Connection reset by peer
[01:15] --> Mellified_Man joined #scummvm.
[01:21] <snover> wjp: for this problem where users in non-english localisations cant upload files to trac, the solution (until Trac 1.3.2+ which is dumping Genshi for jinja2) is apparently to downgrade to genshi 0.6. :\ (https://trac.edgewall.org/ticket/11184)
[01:21] <snover> oh, wowit is actually in their FAQ. really, guys? https://trac.edgewall.org/wiki/TracFaq#Q:WhydoIseeIndexError:popfromemptylistwhentryingtoaddanattachment
[01:35] <snover> id make the changes myself but i dont know how this gets deployed at the moment and dont want to screw anything up :)
[01:45] --> tsoliman__ joined #scummvm.
[01:46] <-- tsoliman__ left irc: Client Quit
[01:49] <-- HaBe left irc: Remote host closed the connection
[01:49] --> HaBe joined #scummvm.
[03:15] <Deledrius> yeah, the whole situation sucks :(
[03:15] <Deledrius> Regardless of format affinity. The lack of support is unfortunate.
[03:50] <-- dreammaster left irc:
[04:09] --> Joefish joined #scummvm.
[04:09] #scummvm: mode change '+v Joefish' by ChanServ!ChanServ@services.
[04:22] <snover> well, the good news is that rather than taking 100 lines of code, writing a PNG file takes only 42.
[04:58] --> Strangerke joined #scummvm.
[05:15] --> Strangerke_ joined #scummvm.
[05:18] <-- Strangerke left irc: Ping timeout: 268 seconds
[05:18] Nick change: Strangerke_ -> Strangerke
[05:23] --> _sev_ joined #scummvm.
[05:23] #scummvm: mode change '+o _sev_' by ChanServ!ChanServ@services.
[05:25] <-- _sev left irc: Ping timeout: 260 seconds
[05:50] --> Begasus joined #scummvm.
[05:53] --> Begas_VBox joined #scummvm.
[06:15] --> ny00123 joined #scummvm.
[06:15] <Deledrius> ^_^
[06:15] <Deledrius> It was pretty simple when I added it to Plasma
[06:21] --> m_kiewitz joined #scummvm.
[06:21] <-- m_kiewitz left irc: Changing host
[06:21] --> m_kiewitz joined #scummvm.
[06:21] #scummvm: mode change '+o m_kiewitz' by ChanServ!ChanServ@services.
[06:39] --> waltervn joined #scummvm.
[06:39] #scummvm: mode change '+o waltervn' by ChanServ!ChanServ@services.
[06:55] <-- Strangerke left irc: Ping timeout: 268 seconds
[07:18] --> ny00123_ joined #scummvm.
[07:21] <-- ny00123 left irc: Ping timeout: 240 seconds
[07:25] <-- LittleToonCat left irc: Remote host closed the connection
[07:35] <-- ny00123_ left irc: Ping timeout: 240 seconds
[07:44] --> Mia joined #scummvm.
[07:50] --> ny00123_ joined #scummvm.
[08:00] <-- Begasus left irc: Ping timeout: 258 seconds
[08:04] --> ajax16384 joined #scummvm.
[08:04] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[08:12] --> Begasus joined #scummvm.
[08:19] --> _sev|work_ joined #scummvm.
[08:19] <-- _sev|work_ left irc: Changing host
[08:19] --> _sev|work_ joined #scummvm.
[08:19] #scummvm: mode change '+o _sev|work_' by ChanServ!ChanServ@services.
[08:20] <_sev|work_> hmmm
[08:20] Nick change: _sev|work_ -> _sev|work
[08:30] <-- _sev|work left irc: Quit: This computer has gone to sleep
[08:33] --> _sev|work joined #scummvm.
[08:33] <-- _sev|work left irc: Changing host
[08:33] --> _sev|work joined #scummvm.
[08:33] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[08:48] --> nutron joined #scummvm.
[09:08] <-- Mia left irc: Ping timeout: 268 seconds
[09:11] --> Mia joined #scummvm.
[09:11] <-- Mia left irc: Changing host
[09:11] --> Mia joined #scummvm.
[09:29] --> t0by joined #scummvm.
[09:29] #scummvm: mode change '+o t0by' by ChanServ!ChanServ@services.
[09:33] <-- _sev|work left irc: Read error: Connection reset by peer
[09:33] --> _sev|work joined #scummvm.
[09:33] <-- _sev|work left irc: Changing host
[09:33] --> _sev|work joined #scummvm.
[09:33] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[10:45] <-- user9 left irc: Ping timeout: 268 seconds
[11:49] --> user9 joined #scummvm.
[12:02] --> omer_mor_ joined #scummvm.
[12:04] <-- omer_mor left irc: Ping timeout: 260 seconds
[12:16] <-- _sev|work left irc: Quit: This computer has gone to sleep
[12:19] --> _sev|work joined #scummvm.
[12:19] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[13:08] <-- ny00123_ left irc: Quit: Leaving
[13:08] <-- _sev|work left irc: Quit: This computer has gone to sleep
[13:11] --> _sev|work joined #scummvm.
[13:11] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[13:12] <-- wanwan left irc: Ping timeout: 260 seconds
[13:18] --> wanwan joined #scummvm.
[13:22] --> HaBe_ joined #scummvm.
[13:27] <-- HaBe left irc: Ping timeout: 272 seconds
[13:52] <m_kiewitz> PSA: Dungeons 2 is free atm on Steam
[13:57] <wjp> humble
[14:01] <m_kiewitz> right
[14:02] <m_kiewitz> DiRT 3 Complete Edition also free via gamesessions.com
[14:03] --> Henke37 joined #scummvm.
[14:20] <-- tsoliman left irc: Ping timeout: 240 seconds
[14:21] --> tsoliman joined #scummvm.
[14:21] #scummvm: mode change '+o tsoliman' by ChanServ!ChanServ@services.
[14:23] <-- tsoliman left irc: Client Quit
[14:23] --> tsoliman joined #scummvm.
[14:23] #scummvm: mode change '+o tsoliman' by ChanServ!ChanServ@services.
[14:23] <-- tsoliman left irc: Remote host closed the connection
[14:24] --> tsoliman joined #scummvm.
[14:24] #scummvm: mode change '+o tsoliman' by ChanServ!ChanServ@services.
[15:04] <m_kiewitz> snover: we finally figured out that IE 11 issue
[15:04] <m_kiewitz> guess what
[15:04] <m_kiewitz> IE 11 goes crazy when you encode a "-" character in hex
[15:05] <logix> OF COURSE, SO OBVIOUS!
[15:05] <m_kiewitz> i replaced "-" with "−", so that the SAP-GUI won't escape it to hex and now it works fine
[15:06] <m_kiewitz> bonus twist: when you put that html data inside a file and then load that file inside IE11, it still takes ages to render
[15:06] <t0by> o_O
[15:07] <logix> why does this sap gui encode a minus char anyway?
[15:07] <m_kiewitz> that's a very good question :P
[15:07] <m_kiewitz> they only do that in the latest version
[15:07] <m_kiewitz> i have no idea what they are doing and they even support IE11 only
[15:08] <m_kiewitz> it's completely weird, there are a few escapes done on application server side, then it's sent to client and then the sap-gui adds even more escapes
[15:08] <logix> "what, it's too slow on your system? buy the new version, we made that faster!"
[15:08] <m_kiewitz> they also escape
[15:08] <m_kiewitz> space
[15:09] <m_kiewitz> brackets and other characters
[15:09] <m_kiewitz> the html looks absolutely crazy
[15:10] <m_kiewitz> there actually is no new version and a regular table takes like 40-60 seconds to render
[15:10] <m_kiewitz> after that fix it renders basically instantly
[15:10] <m_kiewitz> i would even like to look at that IE11 code
[15:10] <logix> last time I saw stuff like that I most probably opened a virus mail to pick apart what it actually was doing
[15:10] <m_kiewitz> sap-gui escapes tons of other characters - all to hex - and everything else works fine
[15:11] <m_kiewitz> it's just the "-" character
[15:14] <logix> https://answers.microsoft.com/en-us/ie/forum/ie11-windows_7/typing-minus-sign-causes-a-crash-after-installing/a0f22437-c8a6-49c1-b154-8eeb86e81f42
[15:14] <logix> probably just a coincidence :)
[15:15] <logix> http://stackoverflow.com/questions/22337498/why-does-ie11-handle-node-normalize-incorrectly-for-the-minus-symbol
[15:15] <m_kiewitz> lol
[15:16] <-- Paaris left irc: Ping timeout: 264 seconds
[15:16] --> Paaris joined #scummvm.
[15:24] <-- Begasus left irc: Ping timeout: 240 seconds
[15:32] --> WooShell joined #scummvm.
[15:34] <WooShell> meow =^.^=
[15:37] --> Begasus joined #scummvm.
[15:55] <-- ajax16384 left irc: Read error: Connection reset by peer
[15:56] <-- Paaris left irc: Ping timeout: 264 seconds
[15:56] --> Paaris joined #scummvm.
[16:11] <Joefish> is there a preferred style for constants? In the coding conventions it says both CONSTANT_NAME and kConstantName are fine but should be unified some day. I looked at 'newer' engines but they also use both styles interchangeably. Is there consensus on what should be used?
[16:18] <m_kiewitz> Joefish: i think kConstantName is the preferred one, at least several of those all-uppercase ones were replaced with kConstantName
[16:18] <m_kiewitz> i personally did prefer the all-uppercase one though :P
[16:23] <logix> just use kCONSTANT_NAME
[16:23] <logix> problem solved
[16:23] <Joefish> m_kiewitz: yea me too :) Of course my attachment doesn't go as far as I would argue for it. Just think having a unified way is much nicer for reading
[16:24] <Joefish> lol
[16:26] <m_kiewitz> yes, i agree
[16:28] <-- _sev|work left irc: Quit: This computer has gone to sleep
[16:33] --> LittleToonCat joined #scummvm.
[16:41] <m_kiewitz> :(
[16:48] <-- Vel0city left irc: Ping timeout: 245 seconds
[16:58] --> SylvainTV joined #scummvm.
[16:58] #scummvm: mode change '+o SylvainTV' by ChanServ!ChanServ@services.
[17:14] --> Vel0city joined #scummvm.
[17:34] --> Farmboy0 joined #scummvm.
[18:10] <wjp> snover: I downgraded genshi to 0.6, increased max_preview_size and enabled ignore_missing_pages
[18:11] <snover> wjp: thanks! :)
[18:23] --> m-kiewitz joined #scummvm.
[18:23] #scummvm: mode change '+o m-kiewitz' by ChanServ!ChanServ@services.
[18:25] <-- m_kiewitz left irc: Ping timeout: 240 seconds
[18:42] <m-kiewitz> snover: have you read what I wrote about IE11 today?
[18:43] --> ajax16384 joined #scummvm.
[18:43] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[18:46] <snover> m-kiewitz: yes
[18:54] --> omer_mor joined #scummvm.
[18:56] <-- omer_mor_ left irc: Ping timeout: 268 seconds
[19:05] <-- t0by left irc: Quit: t0by
[19:10] --> girafe joined #scummvm.
[19:44] <criezy> m-kiewitz, snover: re saving games from GMM in SCI, is it possible to know when the game can be saved safely?
[19:44] <criezy> If not then I guess keeping it disabled is the best options, but otherwise you can enable it and implement canSaveGameStateCurrently() in the engine.
[19:44] <snover> apparently not in sci16, its unclear about sci32
[19:45] <m-kiewitz> no, you can't check for that
[19:45] <m-kiewitz> we tried for ages and there is basically no way to know
[19:45] <m-kiewitz> everything is controlled by the scripts, so unless you figure out every game and every room, it's not possible
[19:45] <m-kiewitz> saving + restoring is fully controlled by scripts
[19:46] <m-kiewitz> 1 solution to this though would be to implement VM saving/restoring (without going through SCI save/restore process)
[19:46] <m-kiewitz> (saving the whole state of the SCI engine)
[19:51] <criezy> I'm not sure it's worth it, unless saving is accessed through some not very accessible ways in some games. But with all the SCI games I have played this is accessed through some buttons in the game, so is easy even on touch devices.
[19:51] <criezy> I implemented GMM load/save for drascula recently only after finding out that saving the game when playing on iPad was difficult since you need to use F11 to access the save dialog.
[19:51] <criezy> And there is no Fn keys on the iOS virtual keyboard and accessing the F# key is done by pressing the key just below the corresponding number key, but there is obviously no '11' key either :P
[19:51] <m-kiewitz> yes, i don't see a real reason for it as well
[19:51] <m-kiewitz> sure, it would be nice
[19:51] <m-kiewitz> but it's a nightmare to do
[19:52] <m-kiewitz> and any small change could break saved game compatibility or would make it difficult to adjust to it
[19:52] <m-kiewitz> and we already offer better restore game options (which is the most important thing imo)
[19:53] <m-kiewitz> like restoring right from ScummVM menu
[19:53] <m-kiewitz> which lets you skip some copy protections and intros for some games as soon as you saw/completed those once
[19:54] <m-kiewitz> and even that took quite some time to get right
[19:54] <m-kiewitz> i remember some games crashing when we did that
[19:55] --> omer_mor_ joined #scummvm.
[19:55] <m-kiewitz> we basically reset part of the VM right before restoring from ScummVM menu, but if I remember right we also start the game for 1 VM loop or so, otherwise there would be other problems
[19:57] <-- omer_mor left irc: Ping timeout: 240 seconds
[20:05] <snover> ugh. i am having a linker problem. this code compiles and links fine in apple clang 7.3.0 on macos, but gcc 6 on debian complains about being unable to find `Graphics::PixelFormat::toString() const` in my new `Image::writePNG` function. libgraphics.a is passed to the compiler before libimage.a, and there is an entry in the symbol table of libgraphics.a for it.
[20:06] <snover> i do not know what stupid thing i have done.
[20:06] --> omer_mor joined #scummvm.
[20:08] <snover> passed to the linker* sorry
[20:08] <-- omer_mor_ left irc: Ping timeout: 240 seconds
[20:09] <snover> this is on my png-screenshots branch for anyone following along at home.
[20:15] <criezy> snover: ld use defined symbols from a object file (or static library) to resolved symbols from object files (or static library) passed before it I think.
[20:15] <criezy> So if the symbol is defined in libgraphics.a, then libgraphics.a needs to be passed after whaterver other .o or .a uses this symbol.
[20:15] <criezy> And it seems to be the other way around in your case?
[20:16] <snover> oh, ok. i guess i remembered the order dependency in&the wrong order :)
[20:16] <snover> according to the output from `make -n` both of them are getting libgraphics.a first
[20:18] <snover> clang just seems to be able to deal with it, i guess. i guess this order comes from Makefile.common
[20:18] <snover> swapping them there fixes the problem. i was so close&
[20:19] <snover> thank you for your help :) i will need it more soon too since i have no confidence that parts of this code will run correctly on big-endian platforms
[20:36] <wjp> I hope swapping them doesn't cause problems in other cases
[20:46] <snover> graphics/macgui/macwindow.cpp might be a problem. it seems like there are a lot of solutions to this, from just passing the libraries multiple times, to using -( -), to using -Wl,--start-group/-Wl,--end-group
[20:47] <m-kiewitz> is there no 2 pass mode available?
[20:47] <m-kiewitz> where it first scans through all files?
[20:47] <wjp> the start/end ones have been problematic in the past, but maybe that's no longer an issue now
[21:03] <snover> im going to open this as a PR just to see what happens on travis, and to get some early feedback
[21:09] --> GitHub4 joined #scummvm.
[21:09] <GitHub4> [scummvm] csnover opened pull request #948: BACKENDS: Compress screenshots using PNG if available (master...png-screenshots) https://git.io/v9Hxh
[21:09] GitHub4 (GitHub4@192.30.252.42) left #scummvm.
[21:16] <wjp> looks like my G4 Mac Mini still boots
[21:17] --> t0by joined #scummvm.
[21:17] #scummvm: mode change '+o t0by' by ChanServ!ChanServ@services.
[21:19] <wjp> but it's always a shock to be reminded of how slow it is :-)
[21:19] <t0by> _sev_, others: sorry for no Director events PR yet, I am slaving over an imminent deadline, will reprise as soon as that's past.
[21:20] <-- t0by left irc: Client Quit
[21:27] --> t0by joined #scummvm.
[21:27] <-- t0by left irc: Changing host
[21:27] --> t0by joined #scummvm.
[21:27] #scummvm: mode change '+o t0by' by ChanServ!ChanServ@services.
[21:29] <wjp> snover: from BE: https://www.usecode.org/scummvm/scummvm00000.png
[21:44] <m-kiewitz> 78kb?
[21:44] <m-kiewitz> o_O
[21:47] <m-kiewitz> even a 320x200 gif is 62k, weird. it should be just 64000 pixels
[21:48] <m-kiewitz> (sure, upscaled, but that's simply straight upscaled)
[21:51] <m-kiewitz> wth, when i save it with paint shop pro, it gets 2k smaller, but when I resize it to 50%, so 320x220, it gets larger - 79k
[21:54] <m-kiewitz> crazy, when I save that 320x220 pic as a RLE compressed bitmap, it's actually smaller than the current output - 68k "only"
[21:56] <-- Henke37 left irc: Quit: ERR_SHUTDOWN
[22:03] <wjp> optipng gets it down to 48514 bytes
[22:04] <-- waltervn left irc: Quit: Leaving
[22:05] <logix> wjp: is that G4 mac mini mostly for BE testing? have you tried pearpc?
[22:05] <m-kiewitz> ah, is there a way to get that into libpng somehow?
[22:05] <m-kiewitz> that's what i would have expected in the first place :/
[22:06] <-- ajax16384 left irc: Read error: Connection reset by peer
[22:08] <wjp> logix: it is, and not recently
[22:08] --> wanwan1 joined #scummvm.
[22:10] <logix> wjp: ok - I can't really say anything about it, I know it exists and that's pretty much it
[22:10] <-- wanwan left irc: Ping timeout: 255 seconds
[22:10] <logix> have been wanting to give it a try for a while...
[22:11] <-- girafe left irc: Read error: Connection reset by peer
[22:18] --> Strangerke joined #scummvm.
[22:21] --> Strangerke_ joined #scummvm.
[22:23] <-- Strangerke left irc: Ping timeout: 268 seconds
[22:23] Nick change: Strangerke_ -> Strangerke
[22:31] <snover> m-kiewitz: the compressor runs on the main thread and the last time i checked OSystem has no way to spawn threads (except for timers), so trying to max out the compression will have a negative effect on user experience.
[22:32] <snover> wjp: glad to hear that it works the first time on BE :)
[22:34] <m-kiewitz> snover: is it so CPU intensive?
[22:34] <m-kiewitz> gif compression takes almost nothing
[22:36] <wjp> given that we've been on bmp for ages without serious complaints, I doubt there's much call for reducing the size further
[22:36] <wjp> optipng basically brute-forces over a whole bunch of compression parameters and picks the ones that lead to the smallest image
[22:36] <wjp> and it reduced it to indexed from rgb
[22:37] <wjp> (lossless of course)
[22:37] <m-kiewitz> i would expect compression to be able to figure out that for example the pic above uses around 180 colors and compress accordingly
[22:38] <m-kiewitz> wouldn't it make sense to check what optipng uses for these
[22:38] <m-kiewitz> and then maybe use those compression parms for 256 color games?
[22:39] <m-kiewitz> or is that a really difficult thing to do? i would simply be interested in how much it improves size in general
[22:40] <m-kiewitz> i also wonder if it would make sense to have 2 screenshot buttons or maybe even 3
[22:40] <m-kiewitz> maybe one for making a screenshot without mouse cursor
[22:41] <m-kiewitz> and maybe another one for making one with original game resolution
[22:42] <m-kiewitz> those could also be debugger commands, but it would be a nice thing to have
[22:53] <-- WooShell left irc: Quit: If you understand or if you don't, if you believe or if you doubt - There's a universal justice, and the eyes of truth are always watching you.
[22:55] <-- Dominus left irc: Read error: Connection reset by peer
[22:55] --> Dominus joined #scummvm.
[23:02] <-- HaBe_ left irc: Quit: Leaving
[23:05] <snover> the amount of time it takes for the compression to run depends upon the size and complexity of the image being compressed (as well as whatever the compression settings are). so, doing some random testing, i have seen times as low as 7ms (running Ms Astro Chicken at 1x) and as high as 150ms (the phantasmagoria 2 main menu) on a 2.3GHz Core i7 with libpngs default settings.
[23:13] <-- t0by left irc: Quit: t0by
[23:26] <m-kiewitz> snover: wow 150ms for 640x400/440/480
[23:27] <m-kiewitz> ah wait, have you turned optimizations on? or is it a debug build?
[23:44] <snover> i thought i was on -O2, but apparently i was not actually on -O2.
[23:47] <snover> only about a 16% speedup though
[23:47] <-- Joefish left irc: Ping timeout: 260 seconds
[00:00] --- Sat May 13 2017