Ren'Py 6.3.0 Released

[c.s.marshall]

Tried it under WinXP SP2: five times as single task, five times under load (P7 rendering in background), five times with VD Compressing video in background. Not a single flaw ...

Mmmmmhhh ... I know I sometimes have *very* peculiar ideas, but what Codec was used for compressing the MPG originally? I ask because there are some compression formats that get buggy when they themselves are compressed (re-encoded) again, like the old MS 263.

One could try the procedure with differently pre-encoded MPGs and see what happens ...

[PyTom]

This is really a phenomenally annyoing bug. Can more people get involved, letting me know if it works (or doesn't) on your machine?

[J13]

Using mpgtest-new.zip and the linked debugger program, Ren'Py crashed when the video was supposed to play and i got the following.

ModLoad: 02b30000 02c29000 C:\WINDOWS\system32\nvwimg.dll
(890.a54): Access violation - code c0000005 (first chance)
First chance exceptions are reported before any exception handling.
This exception may be expected and handled.
eax=016754f0 ebx=008cc9a0 ecx=00000003 edx=00000000 esi=00000000 edi=016754f0
eip=1e015d9a esp=0328fa38 ebp=00000003 iopl=0 nv up ei pl zr na pe nc
cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00010246
*** WARNING: Unable to verify checksum for C:\Desk files\Self productions\mpgtest-new\mpgtest-new\PYTHON23.DLL
*** ERROR: Symbol file could not be found. Defaulted to export symbols for C:\Desk files\Self productions\mpgtest-new\mpgtest-new\PYTHON23.DLL -
PYTHON23!PyEval_EvalCodeEx+0x93a:
1e015d9a 8b4e2c mov ecx,dword ptr [esi+2Ch] ds:0023:0000002c=????????

[Counter Arts]

My machine seems to be working fine all times with the new one... let me test the old one.


EDIT: Tried 3 versions three times each or more. No problems whatsoever!

Man... what are the specs on everyone's machines?

P4 2.4 GHz machine, 512 RAM, winXP SP2

[PyTom]

Okay, added another fix. Hopefully, this one will do it.

[Nafai]

Worked without a glitch for me, except when I tried moving the window around while the video was playing (which resulted in the pixelization earlier reported) but leaving the window alone the movie ran without a hitch.

[J13]

I unfortunately still get the crash with the new version, this is what i got with the debugger, it only ever crashes when the movies supposed to play through a second time though.

ModLoad: 02b30000 02c29000 C:\WINDOWS\system32\nvwimg.dll
(fc8.20c): Access violation - code c0000005 (first chance)
First chance exceptions are reported before any exception handling.
This exception may be expected and handled.
eax=01682750 ebx=008b7160 ecx=00000003 edx=00000000 esi=00000000 edi=01682750
eip=1e015d9a esp=0328fa38 ebp=00000003 iopl=0 nv up ei pl zr na pe nc
cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00010246
*** WARNING: Unable to verify checksum for C:\Desk files\Self productions\mpgtest-new2\mpgtest-new2\PYTHON23.DLL
*** ERROR: Symbol file could not be found. Defaulted to export symbols for C:\Desk files\Self productions\mpgtest-new2\mpgtest-new2\PYTHON23.DLL -
PYTHON23!PyEval_EvalCodeEx+0x93a:
1e015d9a 8b4e2c mov ecx,dword ptr [esi+2Ch] ds:0023:0000002c=????????

and in case specs are relevant
Pentium 4 D 3 ghz
1.5 gb DDR2 Ram
GeForce 7950 GX2

[EwanG]

PyTom,
Seems like the best fix for now might be to just drop support for cutscene movies. I presume there are other ways to do animations for folks who really need them, and I am certain I could rewrite either of my current projects to not use them.

Just a thought,
Ewan

[Enerccio]

PyTom,
Seems like the best fix for now might be to just drop support for cutscene movies. I presume there are other ways to do animations for folks who really need them, and I am certain I could rewrite either of my current projects to not use them.

Just a thought,
Ewan

I hope not though

[Enerccio]

Okay, added another fix. Hopefully, this one will do it.

working 100% here

[chronoluminaire]

Test Version 2 doesn't crash for me, but it does glitch. The first time through it froze on the first screen and didn't animate at all; the second time it glitched for a second or so, then animated nicely. Third time also glitched for a second. Fourth time and fifth times were smooth and fine.

Update: sixth time it crashed and froze

This is on an Athlon dual-core 2.21GHz with 2GB of RAM, running WinXP SP2.

Running it in the debugger, it crashed first time. The command window shows:
(e10.dd8): Access violation - code c0000005 (first chance)
First chance exceptions are reported before any exception handling.
This exception may be expected and handled.
eax=fffffffe ebx=fffffffe ecx=00000000 edx=76b601c0 esi=fffffffe edi=0015c860
eip=76b44fa9 esp=0012f28c ebp=0012f2b4 iopl=0 nv up ei pl zr na pe nc
cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00010246
*** ERROR: Symbol file could not be found. Defaulted to export symbols for C:\WINNT\system32\WINMM.DLL -
WINMM!timeGetTime+0x14e:
76b44fa9 8b40d4 mov eax,dword ptr [eax-2Ch] ds:0023:ffffffd2=????????

The call stack looked like:

WARNING: Stack unwind information not available. Following frames may be wrong.
WINMM!timeGetTime+0x14e
WINMM!midiOutGetDevCapsW+0xf9
WINMM!midiOutGetDevCapsA+0x55
WINMM!midiSteamOpen+0xea
PYTHON23!PyCFunction_Call+0x5a
PYTHON23!PyEval_GetFuncDeck+0x11b
PYTHON23!PyEval_EvalCode+0x1f55

[DaFool]

Is it really crucial to have mpegs obfuscated, given that

1.) JP commercial VNs usually have the mpegs (OPs) unobfuscated
2.) There is already code to unarchive resources from Ren'Py games.

(I would test this, but sadly I'm on a really slow machine right now.).

[Enerccio]

Is it really crucial to have mpegs obfuscated, given that

1.) JP commercial VNs usually have the mpegs (OPs) unobfuscated
2.) There is already code to unarchive resources from Ren'Py games.

(I would test this, but sadly I'm on a really slow machine right now.).

1) Ewan said that he should stop implementing mpgs at all, which is not good
2) Yeah, I don't mind them unobfuscated
3) you cannot use that code to get files from any archive, you can only if you know the name and path of file.

[PyTom]

I'd really like this to work, so I'm spending some time on it.

Hm... chrono's stack trace points to it maybe being a problem with reinitializing the sound system. Could the people that have been having problems please try replacing renpy/audio/audio.py with the attached file?

Thanks.

(I see the corruption that occurs if you move the window around, but for now I'm treating that as a "don't do that" sort of situation.)

[chronoluminaire]

Using that new audio.py, it doesn't glitch at all. First two times played fine.

Third time crashed with identical stack trace and command history to Drake's, as far as I can tell:
WARNING: Stack unwind information not available. Following frames may be wrong.
PYTHON23!PyEval_EvalCodeEx+0x93a
PYTHON23!PyEval_EvalCode+0x25e0
PYTHON23!PyEval_EvalCodeEx+0x70f

(b3c.554): Access violation - code c0000005 (first chance)
First chance exceptions are reported before any exception handling.
This exception may be expected and handled.
eax=01686120 ebx=008d3840 ecx=00000003 edx=00000000 esi=00000000 edi=01686120
eip=1e015d9a esp=0310fa38 ebp=00000003 iopl=0 nv up ei pl zr na pe nc
cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00010246
*** WARNING: Unable to verify checksum for C:\TEMP\mpgtest-new2\PYTHON23.DLL
*** ERROR: Symbol file could not be found. Defaulted to export symbols for C:\TEMP\mpgtest-new2\PYTHON23.DLL -
PYTHON23!PyEval_EvalCodeEx+0x93a:
1e015d9a 8b4e2c mov ecx,dword ptr [esi+2Ch] ds:0023:0000002c=????????


Update: I tried restarting the debugger. This time it glitched again the first time, and the second time froze, without any stack trace or info in the command window. That precise sequence happened twice more.

Looks like you've fixed the audio problem I was having, but the main problem of Drake's and others' is still around.