forums.marpirc.net

Hard to believe it's been two and a half years since this thread:

http://www.marpirc.net/viewtopic.php?f= ... 62e759b924

And I finally tried my hand at building an xwolfmame 106 that actually does the same recording as the Twin Galaxies version. Of course it's up to TG if they would accept submissions from it, but I've done my best to make it byte-for-byte indistinguishable from the Windows recordings, including spoofing Windows XP OS info. The results pass wlfview and an extended wlfview that I wrote. I think people will notice the attention to detail, however it hasn't been tested much, since I have limited access to Windows machines (hence my motivation). Non-Intel CPUs might not produce an identical recording, since I did not have Windows running on AMD to try. Also, I've checked a lot of options, but maybe there are some that are disabled in the TG version that would show up in this .wlf file. Maybe people could tell me what they often use, and I'll try them out. If there is something mameplus specific for a particular game, that could be an issue, since this was build against xmame-0.106. I couldn't find the released mameplus 106 source anywhere.

The ultimate goal is to have your recordings play to completion in TG's xwolfmame106, and I expect they will for the vast majority.

I'd love to get this working on Mac too, but I'm not sure how to proceed. Mame 141 compiles so easily on Snow Leopard, it seems like a waste to work on old code. I also haven't played mame in a year and I've got some gaming to catch up on.

I can't upload the 9.3MB zipped binary, so imagine if you will that such a version was out there. Would you use it?

http://wolfmame.marpirc.net/xwolfmame-0106.tar.bz2 -- Uploaded back in 2008, but it's really only designed for playback. Recording is possible, but can be problematic. More recent SDL versions are recommended, and a gazillion times better than crusty old X-MAME.

mahlemiut wrote:http://wolfmame.marpirc.net/xwolfmame-0106.tar.bz2 -- Uploaded back in 2008, but it's really only designed for playback. Recording is possible, but can be problematic. More recent SDL versions are recommended, and a gazillion times better than crusty old X-MAME.

I was on the thread when you posted it in 2008 (see above). The interesting part was supposed to be "TG compatible".

Since WolfMAMEplus 0.106 is required so often, I've always wanted one that records correctly (i.e. TG compatible) for Linux. Neither the mameplus nor the xmame diffs posted at the time recorded the same as wolfmameplus-0106. So, I thought I'd wait for TG to update their MAME version on their own schedule, but time dragged on. Two years later I finally became impatient enough to determine the format by examination and experimentation.

xwolfmame-0.106-TG-0.1.zip

Anyway, it was a big deal for me, since I don't use Windows, but maybe I'm the only one.

The file format is the same across all platforms. The differences between MAME, X-MAME (RIP) and MAME Plus are down to differences in the MAME core, and maybe some slight timing differences. That being said, the majority of all INPs should work in any platform, so long as it's based on 0.106.

It's another good reason for using a more recent version, now that the SDL code is in baseline MAME, and builds out of the box on Windows, Linux, BSD or OS X. You can use the same WolfMAME diffs for whichever platform.

The wlf file version info, options, file size and contents in the available diffs are very different than the longstanding wolfmameplus-106, and thus not accepted at Twin Galaxies. That's the story here. I suppose the INP header is only different in the version number from xwolfmame-104, but since the previous xwolfmame-106 version you linked can't record at all, the comparisons should really end there. This is the recording version - with a valid wlf file, according to wlfview.exe.

If allowed, I would use the latest. I don't disagree.

The issue with wlf files is a matter of variable alignment. More that it differs between the standard Linux version of GCC and the MinGW32 version of GCC. Basically, under Linux, the compiler adds padding between certain variables in a structure so that they are located at a point aligned on a 2, 4 or 8 byte boundary, depending on the variable's size. I did fix it, by setting a #pragma to change the alignment setting to match MinGW32, but I guess I didn't add that bit when quickly whipping up the 0.106 binary.

mahlemiut wrote:The issue with wlf files is a matter of variable alignment. More that it differs between the standard Linux version of GCC and the MinGW32 version of GCC. Basically, under Linux, the compiler adds padding between certain variables in a structure so that they are located at a point aligned on a 2, 4 or 8 byte boundary, depending on the variable's size. I did fix it, by setting a #pragma to change the alignment setting to match MinGW32, but I guess I didn't add that bit when quickly whipping up the 0.106 binary.

Nice. I didn't know the pragma usage for alignment, so I just stuck in a couple of extra variables in there. I'd like to make a couple of points though. There are over 20 options in the wlf file that are not in the diff. Maybe I would have found most of those in the base mameplus source, but neither 106 nor anything near it is on the net anymore. (If you have that tar, BTW, that would be great.) Also, I can see the diff is not exactly TG's version, but it was close enough. I couldn't have done it without that. Thanks.

aycaramba, if you made any source code changes to wolf106 to fix wlf file creation can you make those changes public? I'd also like to have a version of xwolf that records properly at all times.

Try this one, skito.

-ayc