[Back to Index]

[00:03] <wjp> oh dear, this qfg1 palette thing is going to have to be an extremely creative script patch
[00:03] <wjp> need about 30-40 bytes
[00:05] <-- Farmboy0 left irc: Remote host closed the connection
[00:07] <snover> i dont know if it is relevant at this late stage of the game (maybe it is because of QFG4), but i dont know that there is any technical reason why the script patcher cannot track deltas for insertion points and then fix up the object and script offsets using those deltas so its not necessary to be quite so creative.
[00:09] <wjp> you mean doing relocations in the entire script?
[00:09] <snover> yes. or even more simply just tell it to extend the end of the script, dump the extra patch bytes at the end, and insert a jump to it.
[00:11] <wjp> the problem is the relative offsets
[00:11] <wjp> putting things at the end would avoid that of course
[00:12] <-- criezy left irc: Quit: criezy
[00:13] <wjp> in theory we _could_ keep track of all insertion points and fix up relative offsets at run-time when encountering them... but... icky
[00:14] <snover> it would be more perfectionistically satisfying to do that, and way more practical and less of a nightmare to just extend and jump to the end
[00:16] <wjp> so this particular example would only need to wrap two local functions into two new local functions
[00:17] <wjp> (and then replace calls to the old ones with calls to the new ones)
[00:17] <wjp> which would be quite simple if we can just put the new functions at the end
[00:18] <wjp> (and then it will of course turn out that this whole strategy of PalVary calls I have planned won't work...)
[00:19] <snover> yeah. well, like i said, probably still going to be helpful for QFG4, since fixing bugs there by NRS meant recompiling a bunch of scripts since there was not space to fix the bugs properly
[00:19] <snover> bbiaf
[00:19] <wjp> mhm
[00:29] <wjp> there's a bunch of sanity checks for jumping past the end of the script, so putting extra code past the end of the heap won't directly work
[00:30] <snover> wjp: sanity checks where?
[00:30] <wjp> vm
[00:31] <wjp> error("[VM] op_jmp: request to jump past the end of script %d (offset %d, script is %d bytes)",
[00:31] <wjp> and such
[00:31] <wjp> those can presumably all be modified
[00:32] <snover> iirc that would be only relevant for the 1.12.1 though, since there is no separate heap for SCI01 or 3
[00:33] <snover> what i was thinking was that Script::load would query ScriptPatcher its extra script bytes and it would just extend the size of _script by that amount
[00:34] <snover> to the rest of the VM it would look like a normal unaltered script
[00:35] <snover> uint32 scriptSize = script->size() + scriptPatcher->getExtraBytes(_nr)
[00:37] <snover> then maybe for the script patches themselves, some new PATCH_ENDJMP or something to demarcate the stuff that goes at the end so it can create an appropriate relative jump automatically regardless of the rest of the script size
[00:39] <wjp> hrm, I'd need to check how making the script longer will affect pointers into the heap
[00:41] <snover> it shouldnt, IIRC. in SSCI1.1 the heap was just designed as a separate segment, and in our VM we recombine it onto the end of the script memory and those offsets are relative from 0 of the heap
[00:42] <snover> the only problem i can think of offhand would be if someone was reading one of the tables at the end of the script through the end of the script instead of reading the length of the table, but i *think* i fixed all of that when i added Span.
[00:45] <snover> so i am fairly confident that this is will be way more straightforward than anyone would have feared.
[00:46] <snover> -is
[00:47] <wjp> very disappointing
[00:48] <snover> i know, sorry for making something easy for once :)
[00:48] <wjp> there go all my clever tricks to turn a bunch of calls into a while loop to save 40 bytes
[00:50] <wjp> but maybe 1:50am isn't the best time to start on this
[00:51] <wjp> so good night :-)
[00:51] <snover> far too rational. i dont like it!
[00:51] <snover> good night!
[00:51] <snover> :)
[01:13] <tsoliman> snover: https://pastebin.com/uqLB9uK3
[01:14] <snover> curious.
[01:15] <tsoliman> I am gonna pull your new images now
[01:15] <tsoliman> should I just pull android or both?
[01:15] <tsoliman> meaning do I have to pull the buildmaster as well?
[01:17] <snover> no buildmaster changes yet.
[01:18] <-- Strangerke left irc: Ping timeout: 268 seconds
[01:19] <tsoliman> I am loving this buildbot UI (when it isn't acting up)
[01:20] <snover> its better than the old one in a lot of ways, and it is written using coffeescript and angular so is worse than the old one in a lot of ways
[01:22] <tsoliman> I am also a fan of how I just relaunched the worker as a new version and the old build is there still .. all I had to do is press "rebuild" .. the old worker container wasn't the host of the logs and such
[01:23] <snover> yes, the buildmaster has a sqlite database which holds that information and it is persisted into a volume so you can do whatever and it will always come back
[01:23] <tsoliman> same error with the new image .. it is no big deal for now
[01:25] <-- waltervn left irc: Quit: Leaving
[01:25] <snover> its a scary error, those are function-local variables so how they would be undefined randomly is worrisome
[01:48] --> Strangerke joined #scummvm.
[01:55] <-- ccawley2011 left irc: Quit: Page closed
[02:02] --> Littleboy joined #scummvm.
[02:02] <-- Littleboy left irc: Client Quit
[02:11] <wjp> Both arrays are const and equal to another array there
[02:12] <wjp> I'm guessing they have been merged but the generared debugging info is still pointing to the disappeared one
[02:12] <wjp> (Note the .debug_addr segment in the error)
[02:13] <snover> insomnia? :)
[02:14] <wjp> Would be interesting to try if the error disappears if you make the arrays unique
[02:15] <snover> i noted the debug_addr and then dismissed it
[02:15] <snover> what you are saying makes sense
[02:18] <snover> not great, and still, better than something that would actually impact builds in a material way since the debugging optimisations can just be changed/disabled/whatever.
[02:23] <-- Dominus left irc: Ping timeout: 260 seconds
[02:53] <-- dreammaster left irc:
[03:06] <-- Strangerke left irc: Ping timeout: 268 seconds
[03:15] --> Strangerke joined #scummvm.
[03:31] <-- ignalina left irc: Quit: ignalina
[03:34] <snover> yeaayyyyyy macos buildbot works finally.
[03:38] <-- SylvainTV left irc: Read error: Connection reset by peer
[03:54] <-- dafioram left irc: Ping timeout: 255 seconds
[03:57] --> Strangerke_ joined #scummvm.
[04:00] <-- Strangerke left irc: Ping timeout: 248 seconds
[04:00] Nick change: Strangerke_ -> Strangerke
[06:19] <-- Strangerke left irc: Ping timeout: 260 seconds
[07:08] --> Begasus joined #scummvm.
[07:19] --> Begas_VBox joined #scummvm.
[07:54] --> Dominus joined #scummvm.
[08:11] --> jamm joined #scummvm.
[08:17] --> Strangerke joined #scummvm.
[08:56] <-- Begasus left irc: Ping timeout: 260 seconds
[09:08] <-- LittleToonCat left irc: Remote host closed the connection
[09:08] --> Begasus joined #scummvm.
[09:47] --> jam__ joined #scummvm.
[09:47] <-- jamm left irc: Ping timeout: 260 seconds
[09:54] --> ignalina joined #scummvm.
[10:28] --> waltervn joined #scummvm.
[10:28] #scummvm: mode change '+o waltervn' by ChanServ!ChanServ@services.
[10:28] <waltervn> morning
[10:33] <Begas_VBox> hi waltervn
[10:34] <waltervn> hey Luc, how's Haiku doing? I haven't really kept up lately
[10:36] <Begas_VBox> still progressing :)
[10:37] <Begas_VBox> QT is working pretty good, even got some KDE apps running on Haiku now
[10:38] <Begas_VBox> need to check the qxwidgets if the can produce the tools application shortly ;)
[10:39] <waltervn> are you using Haiku as your main OS or is that still too early?
[10:41] <Begas_VBox> still too early, compiling works better for me in VBox then native
[10:46] <logix> ooh, I just wanted to ask if it works in vbox
[10:46] <logix> nice, something to try out some time
[10:47] <Begas_VBox> the more the merior (if I spelled that correctly) ;)
[10:53] <logix> merrier :P
[10:55] <Begas_VBox> thougth so ;)
[11:04] --> ajax16384 joined #scummvm.
[11:04] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[11:08] <logix> what's an easy way to get it up and running, can you download an install iso and boot that in vbox?
[11:08] <waltervn> I think it would have been better to build Haiku on top of the Linux or BSD kernel
[11:08] <logix> Begas_VBox: ignore me, I found https://www.haiku-os.org/guides/virtualizing/virtualbox/
[11:08] <logix> <- mad google skillz
[11:09] <Begas_VBox> ah sorry :)
[11:09] <logix> waltervn: but wasn't the point (or part of the point) to do "something new"?
[11:09] <Begas_VBox> looking into mandb (gdbm atm) :)
[11:10] <Begas_VBox> you can download one of the nightlies (anyboot image), you can mount that one :)
[11:12] <waltervn> logix: what I liked about BeOS was that it provided a consistent (is that the word?) user interface, built on a unix-like base, in contrast to the wild-west you have on Linux.
[11:15] <Begas_VBox> in regard of the wild west, we're getting there too I guess ;)
[11:15] <waltervn> the mention of QT did worry me a bit :P
[11:15] <Begas_VBox> ;)
[11:16] <Begas_VBox> never got wxwidgets to build under BeOS/Haiku ... so maybe with qxwidgets we get a chance for the GUI tools for ScummVM :)
[11:26] <logix> waltervn: "consistent" sounds right, yes
[11:35] <logix> is beos usable in a gui-less way? I mean, is there an sshd and such?
[11:35] <logix> I understand that you'd miss out on a big part of it of course
[11:36] <wjp> waltervn: in our amigamac sci midi driver, instrument->loop_size can be larger than 32k, which means it will overflow a frac_t
[11:36] <logix> ah, yes, https://www.haiku-os.org/guides/daily-tasks/netservices/ mentions sshd
[11:39] <waltervn> wjp: what's the end result when that happens?
[11:40] <wjp> invalid reads from the samples array
[11:41] <wjp> (and a segfault)
[11:41] <waltervn> I'm fine with any possible fix, including the full removal of the driver ;)
[11:44] <waltervn> sorry, there are kids yelling all around me, let me try coming up with a better reponse
[11:45] <wjp> I think I have an easy patch that'll do for now
[11:46] <waltervn> https://github.com/waltervn/scummvm/blob/sci_sound/engines/sci/sound/drivers/amigasci0.cpp and https://github.com/waltervn/scummvm/blob/sci_sound/engines/sci/sound/drivers/amigasci1.cpp are a lot closer to the original drivers, maybe they can provide some information
[11:46] <waltervn> and these: https://github.com/waltervn/scummvm/blob/sci_sound/engines/sci/sound/drivers/macsci0.cpp https://github.com/waltervn/scummvm/blob/sci_sound/engines/sci/sound/drivers/macsci1.cpp
[11:50] --> Strangerke_ joined #scummvm.
[11:51] <wjp> so I won't dive into this too deeply to see if the instrument data is wrong or not, and just commit a quick fix then
[11:52] <-- Strangerke left irc: Ping timeout: 255 seconds
[11:52] Nick change: Strangerke_ -> Strangerke
[11:54] --> GitHub121 joined #scummvm.
[11:54] <GitHub121> [scummvm] wjp pushed 1 new commit to master: https://git.io/vbOLH
[11:54] <GitHub121> scummvm/master 3a93313 Willem Jan Palenstijn: SCI: Avoid buffer overflow in amigamac sound driver...
[11:54] GitHub121 (GitHub121@192.30.252.45) left #scummvm.
[11:54] <wjp> such as that :-)
[11:58] <waltervn> after the MIDI code gets fixed I plan to dive back into the drivers
[11:58] <wjp> that does seem like the right order
[11:59] <waltervn> these new drivers aren't really ready for inclusion. They need additional safety checks and testing. However they *are* closer to the original, and actually produce something similar to music in SCI1 ;)
[11:59] <waltervn> it could be argued that SCI1 mac/amiga games shouldn't be marked as supported
[11:59] <wjp> great :-)
[12:03] <waltervn> see also http://forums.scummvm.org/viewtopic.php?t=14293 for example. In a sense it is unfortunate that these new drivers aren't included yet, but I can't really find the energy to do this until we have the MIDI code fixed.
[12:34] --> mortalius_ joined #scummvm.
[12:35] <-- mortalius left irc: Ping timeout: 260 seconds
[12:43] --> nast joined #scummvm.
[12:43] <-- jam__ left irc: Ping timeout: 255 seconds
[13:01] --> Farmboy0 joined #scummvm.
[13:01] <-- Farmboy0 left irc: Changing host
[13:01] --> Farmboy0 joined #scummvm.
[13:05] <-- nast left irc: Quit: Leaving
[13:08] <-- Strangerke left irc: Ping timeout: 248 seconds
[13:18] --> ccawley2011 joined #scummvm.
[13:38] --> Strangerke joined #scummvm.
[14:11] --> SylvainTV joined #scummvm.
[14:11] #scummvm: mode change '+o SylvainTV' by ChanServ!ChanServ@services.
[14:39] --> jam__ joined #scummvm.
[15:21] --> girafe joined #scummvm.
[15:22] <-- Begasus left irc: Ping timeout: 260 seconds
[15:30] <-- DDR left irc: Ping timeout: 258 seconds
[15:33] <-- eriktorbjorn1 left irc: Quit: Leaving.
[15:34] --> Begasus joined #scummvm.
[15:35] --> eriktorbjorn joined #scummvm.
[15:35] #scummvm: mode change '+o eriktorbjorn' by ChanServ!ChanServ@services.
[15:38] --> DDR joined #scummvm.
[16:16] <-- jam__ left irc: Ping timeout: 255 seconds
[16:24] --> t0by joined #scummvm.
[16:24] #scummvm: mode change '+o t0by' by ChanServ!ChanServ@services.
[16:25] <t0by> Hey
[16:25] <t0by> Quick heads up: http://forums.irix.pw/viewtopic.php?f=10&t=7
[16:25] <t0by> TL;DR: Free IRIX hardware for anybody who volunteers porting stuff.
[16:26] <t0by> Strangerke, you might be particularly interested in your capacity as PR man ^
[16:27] <snover> what unique thing is that one can do with such hardware?
[16:27] <snover> is it*
[16:27] <t0by> Running IRIX and porting stuff to it and perhaps building stuff, I guess.
[16:27] <t0by> I have absolutely no idea what the ScummVM/IRIX situation is currently
[16:28] <t0by> But I seem to remember porters are inactive and/or disappeared
[16:29] <snover> t0by: i recently aggregated scummvms download stats which you may be interested in regarding this. https://docs.google.com/spreadsheets/d/1NBZNK2d7b4Ofasi8SXcgVdHVfKiz82fvn-BK3EtGYDI/edit#gid=817497482
[16:29] <t0by> Believe it or not I was wondering about download stats no earlier than a week ago
[16:30] <t0by> That's super neeeeeat.
[16:33] --> Littleboy joined #scummvm.
[16:33] #scummvm: mode change '+o Littleboy' by ChanServ!ChanServ@services.
[16:33] <-- Littleboy left irc: Client Quit
[16:36] <logix> t0by: dang, I already own an SGI
[16:38] --> ny00123 joined #scummvm.
[16:40] <mal> hi, I was looking at the old PRs on github and this would be quite useful https://github.com/scummvm/scummvm/pull/401
[16:40] <snover> mal: so you are willing to work on that to make it up-to-date so it can be merged?
[16:41] <mal> well I was updating the sailfish port and the touchscreen code in that is quite ugly so probably would be better to get that in first
[16:55] <mal> snover: I'll try to rebase that
[16:57] --> Henke37 joined #scummvm.
[17:02] <snover> mal: it needs also the feedback responded to and addressed
[17:04] <mal> snover: sure
[17:11] <snover> awesome!
[17:12] <mal> rebase went quite easily, now testing if it still builds
[17:25] --> LittleToonCat joined #scummvm.
[17:41] <-- t0by left irc: Quit: Leaving
[17:48] --> TMM joined #scummvm.
[17:48] <-- TMM left irc: Changing host
[17:48] --> TMM joined #scummvm.
[17:48] #scummvm: mode change '+o TMM' by ChanServ!ChanServ@services.
[18:03] --> dreammaster joined #scummvm.
[18:03] #scummvm: mode change '+o dreammaster' by ChanServ!ChanServ@services.
[18:11] --> GitHub13 joined #scummvm.
[18:11] <GitHub13> [scummvm] dreammaster pushed 1 new commit to master: https://git.io/vbORs
[18:11] <GitHub13> scummvm/master 4fbf11d Paul Gilbert: GRAPHICS: Fix ManagedSurface clear calls when no surface is set
[18:11] GitHub13 (GitHub13@192.30.252.35) left #scummvm.
[18:21] <-- Strangerke left irc: Ping timeout: 260 seconds
[18:28] --> Strangerke joined #scummvm.
[18:35] --> GitHub46 joined #scummvm.
[18:35] <GitHub46> [scummvm] csnover pushed 1 new commit to branch-2-0: https://git.io/vbO0D
[18:35] <GitHub46> scummvm/branch-2-0 58cbd45 Colin Snover: SDL: Fix memory corruption when switching to/from 32-bit cursors...
[18:35] GitHub46 (GitHub46@192.30.252.35) left #scummvm.
[18:35] --> GitHub59 joined #scummvm.
[18:35] <GitHub59> [scummvm] criezy pushed 1 new commit to master: https://git.io/vbO0S
[18:35] <GitHub59> scummvm/master 8b1488a Thierry Crozat: I18N: Update translations templates
[18:35] GitHub59 (GitHub59@192.30.252.45) left #scummvm.
[18:40] <snover> dreammaster: if it is possible, could you use the -x switch when cherry-picking from master to the release branch, so that the commits picked from master are cross-referenced in the commit message?
[18:41] <dreammaster> I'm not familiar with the option, but I'll try to remember if anything else comes up
[18:41] <snover> it just makes it a bit easier to verify at a glance that something was pulled into the branch from master and didnt get committed only to the branch.
[18:42] <-- Strangerke left irc: Ping timeout: 240 seconds
[18:42] <snover> thanks!
[18:42] <dreammaster> NP
[18:43] --> Strangerke joined #scummvm.
[18:47] <mal> snover: I modified the touchmapper code in that PR so that it works even without touchscreen, now it has a fallback to defaultmapper for other events so for example sdl backend seems to work fine on my laptop with touchmapper enabled
[18:48] --> Axy joined #scummvm.
[18:48] <-- Axy left irc: Changing host
[18:48] --> Axy joined #scummvm.
[18:49] <snover> mal: ok. i dont really know anything about the content of the PR, i just reviewed all the open PRs a month or two ago and it had issues which prevented it from being landed, so that sounds good, and i have no idea what it actually means for the end-result of the PR :)
[18:49] <-- Mia left irc: Read error: Connection reset by peer
[18:49] <mal> snover: we'll see how it looks after some cleanup
[18:53] --> Mia joined #scummvm.
[18:53] <-- Mia left irc: Changing host
[18:53] --> Mia joined #scummvm.
[18:56] <-- Axy left irc: Ping timeout: 248 seconds
[18:59] <-- Strangerke left irc: Ping timeout: 260 seconds
[18:59] <snover> looking forward to it!
[19:11] --> Strangerke joined #scummvm.
[19:21] <-- TMM left irc: Quit: Ex-Chat
[19:34] <-- Begasus left irc: Ping timeout: 264 seconds
[19:37] --> TMM joined #scummvm.
[19:37] <-- TMM left irc: Changing host
[19:37] --> TMM joined #scummvm.
[19:37] #scummvm: mode change '+o TMM' by ChanServ!ChanServ@services.
[19:45] --> Begasus joined #scummvm.
[19:55] <-- girafe left irc: Quit: Leaving
[19:59] --> girafe joined #scummvm.
[20:06] <-- Begasus left irc: Ping timeout: 260 seconds
[20:18] --> Begasus joined #scummvm.
[20:43] <-- dreammaster left irc:
[20:44] <-- ajax16384 left irc: Read error: Connection reset by peer
[20:50] <-- Strangerke left irc: Ping timeout: 260 seconds
[21:04] --> Strangerke joined #scummvm.
[21:07] --> Strangerke_ joined #scummvm.
[21:10] <-- Strangerke left irc: Ping timeout: 260 seconds
[21:10] Nick change: Strangerke_ -> Strangerke
[21:22] <-- Begas_VBox left irc: Quit: Vision[0.10.2]: i've been blurred!
[21:27] <-- Begasus left irc: Quit: Ex-Chat
[21:32] #scummvm: mode change '+o Strangerke' by ChanServ!ChanServ@services.
[21:37] <Strangerke> hi guys
[21:37] <Strangerke> tsk
[21:37] --> dreammaster joined #scummvm.
[21:37] #scummvm: mode change '+o dreammaster' by ChanServ!ChanServ@services.
[21:47] <-- macdude22 left irc: Quit: Textual IRC Client: www.textualapp.com
[22:08] --> GitHub58 joined #scummvm.
[22:08] <GitHub58> [scummvm] sev- pushed 6 new commits to branch-2-0: https://git.io/vbO64
[22:08] <GitHub58> scummvm/branch-2-0 a08ebc2 Eugene Sandulenko: FULLPIPE: Fix bug #9672: "Several inventory items highlighted at the same time"
[22:08] <GitHub58> scummvm/branch-2-0 6bf5b1a Eugene Sandulenko: FULLPIPE: Fix bug #9673: Unable to move after loading a save from the title screen
[22:08] <GitHub58> scummvm/branch-2-0 c0afab5 Eugene Sandulenko: FULLPIPE: Fix bug #10325: New game does not put back already picked up items
[22:08] GitHub58 (GitHub58@192.30.252.45) left #scummvm.
[22:12] <snover> wjp: so as it turns out, i cant actually test the iOS builds in iOS simulator because i forgot that in the simulator, apps are all actually compiled for x86_64.
[22:13] <snover> i dont really know how useful this is in any case, we should really just try to send to the app store.
[22:36] <wjp> surely there must be a way for devs to test applications on a real device?
[22:37] <snover> yes, you may connect a real device over USB and then have Xcode build, sign, and install it to the device
[22:51] <-- dreammaster left irc:
[22:52] <-- Henke37 left irc: Quit: ERR_SHUTDOWN
[22:53] --> dreammaster joined #scummvm.
[22:53] #scummvm: mode change '+o dreammaster' by ChanServ!ChanServ@services.
[22:57] <wjp> do we _have_ anyone with a real device?
[22:59] <logix> I got an iphone 3 w/ ios 3.1.2 :P
[22:59] <wjp> the way you say that makes me suspect that's not very recent?
[22:59] <snover> i have an ipad. its not jailbroken. ill have to check how/if i can codesign and load the binary generated by buildbot instead of one from xcode.
[23:00] <logix> wjp: I bought it 2nd hand in 2011
[23:00] <snover> iOS 3 was released in 2009 and superseded by 2010
[23:00] <snover> iOS 7.1 is the minimum for the Buildbot builds.
[23:01] <logix> it's more than half the age of the first iphone ever built
[23:01] <wjp> you should be able to run codesign on an own build
[23:01] <logix> well, the first iphone ever sold I guess...
[23:01] <logix> so - probably not what you'd call very recent, yes
[23:02] <snover> the biggest roadblock for me is probably that i refuse to allow itunes to infect my computer so i cant transfer through that, and i am not sure offhand how else i can transfer over usb.
[23:02] <snover> i imagine there is something, i have just not done any research about it.
[23:02] <snover> also if the build doesnt just work i am not sure if i can get logs without having xcode manage the process, and i can cross that bridge if i come to it.
[23:03] <wjp> or apparently there's also the TestFlight service
[23:05] <-- Lightkey left irc: Ping timeout: 276 seconds
[23:05] <snover> iirc thats more for distributing and managing beta tests and requires passing an apple review
[23:05] <snover> as well as having an actual ios dev account which costs money
[23:06] <wjp> mhm
[23:07] <snover> ill be back in a couple hours.
[23:07] <-- ny00123 left irc: Quit: Leaving
[23:08] <-- rootfather left irc: Ping timeout: 248 seconds
[23:18] --> Lightkey joined #scummvm.
[23:25] <-- girafe left irc: Quit: Leaving
[23:29] --> girafe joined #scummvm.
[23:39] <-- Farmboy0 left irc: Remote host closed the connection
[00:00] --- Mon Dec 4 2017