Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bugfinder]

Hi folks, long time no see. Life's keeping me busy, so I can't find enough time to follow the game emulating scene like the old days, but I try to at least keep my MAME ROM collection complete. That's a compulsion many of us have, and I was never immune to it. Since R. Belmont said a long time ago "MAMEDev can waste your disk space with impunity"... we who suffer from ROM Collection Syndrome can only accept.

Speaking of keeping my collection complete, I must extend my deepest thanks to Roman. His tool ClrMAMEPro has been invaluable since its creation, almost 25 years ago. Today I made a donation, so beer's on me, my friend. Greetings from Brazil.

Now, onto business: I have a complete (minus software lists) MAME 0.268 ROMset in split form. I like the idea of a full merged ROMset, but ClrMAMEPro 4.048d can't do it the way I want, with clone subfolders inside their parents' zipfiles. In merge mode, it will have some clone folders inside a parent zipfile, but not all. With the release of new tools, I decided to try them out.

First thing I did was to copy all ROMsets and CHDs to another drive, so I could preserve what I already had and make comparisons between results. I ran rebuilder 0.09 using the 0.268 split ROMset sanctioned by ClrMAMEPro and it took about 4 hours to complete. The total numer of zipfiles reduced from 41,776 to 14,766 (64.7% less files). It became faster to navigate within a folder with a lot less files (of course, total size remained approximately the same).

But I had my doubts if MAME would agree with the output generated by the new rebuilder. I decided to have MAME verify both collections and them compare results. I got -listxml, filtered all <machine name=> inside of it, which amounts to 47,016 items. Being curious, I discovered that Roman's tools fix the -listxml output, replacing commas by periods in some machines' refresh rates (typos in the source). Then I made a batch file where MAME would verify each "machine" and output verify results to text files of same names. Like this:

mame 005 -verifyroms >005.txt
mame 09825_67907 -verifyroms >09825_67907.txt
mame 100lions -verifyroms >100lions.txt

After MAME finished verifying both collections, I had 47,016 txt files on each drive. 5,323 of them had 0 bytes; that's because these refer to "machines" without ROMS, so MAME couldn't verify them. 575 ROMsets didn't exist as zipfiles in my split ROM collection; I assume it's because they are clones whose ROMs match exactly the same as their parents (like kizuna4p is a clone of kizuna, but kizuna4p.zip isn't created by ClrMAMEPro). The remaining 41,118 txt files were all OK in both drives. That's great!

Side note: I am puzzled as to why some machines which don't have ROMs receive OK while others don't. For example, mame ac97 -listroms shows [No ROMs required for device "ac97"], same with mame a2600 -listroms [No ROMs required for driver "a2600"]. But when verified, MAME says a2600 is OK (best available) while for ac97 it says it has no ROMs. Maybe Roman can explain this apparent double standard? I guess it's because one is a device (ac97) and the other is a driver (a2600), but still.

Continuing, the new rebuilder inserted 13 files marked as NO GOOD DUMP KNOWN inside 11 zipfiles. I manually identified and copied them into my previously existing collection, then had ClrMAMEPro performing its check. It warned me about those 13 files being unneeded:

Link (Korean bootleg of Atari Tetris) [folder: link - parent: atetris - size: 132kb]
unneeded file: D:\mame\roms\link\82s123.bin [not fixed]

Cuore 1 (Italian, set 6) [folder: cuoreunoe - parent: cuoreuno - size: 99kb]
unneeded file: D:\mame\roms\cuoreunoe\palce16v8h-25pc-4.u5 [not fixed]

Cuore 1 (Italian, set 7) [folder: cuoreunof - parent: cuoreuno - size: 99kb]
unneeded file: D:\mame\roms\cuoreunof\palce16v8h-25pc-4.u5 [not fixed]

Diana Bifuca (v9.25) [folder: dibifuca - size: 134kb]
unneeded file: D:\mame\roms\dibifuca\24lc16b.ic8 [not fixed]

Diana Bifuca (v7.43) [folder: dibif743 - parent: dibifuca - size: 134kb]
unneeded file: D:\mame\roms\dibif743\24lc16b.ic8 [not fixed]

Desert Patrol [folder: dpatrol - size: 3kb]
unneeded file: D:\mame\roms\dpatrol\bd2.h7 [not fixed]

Driving Force (Galaxian conversion, Seatongrove UK, E-0237) [folder: drivfrcsga - parent: drivfrcp - size: 32kb]
unneeded file: D:\mame\roms\drivfrcsga\82s123-1.bin [not fixed]
unneeded file: D:\mame\roms\drivfrcsga\82s123-2.bin [not fixed]

Jolly Card (Italian, encrypted bootleg, set 2) [folder: jolycdic - parent: jollycrd - size: 195kb]
unneeded file: D:\mame\roms\jolycdic\gal16v8b.bin [not fixed]

Jolly Card (Italian, encrypted bootleg, set 3) [folder: jolycdii - parent: jollycrd - size: 100kb]
unneeded file: D:\mame\roms\jolycdii\gal16v8b.bin [not fixed]

Lup Lup Puzzle / Zhuan Zhuan Puzzle (version 3.0 / 990128) [folder: luplup - parent: suplup - size: 9mb]
unneeded file: D:\mame\roms\luplup\gal22v10b.gal1 [not fixed]

Hipoly (bootleg of Hyper Olympic) [folder: hipoly - parent: trackfld - size: 137kb]
unneeded file: D:\mame\roms\hipoly\pal16l8.e4 [not fixed]
unneeded file: D:\mame\roms\hipoly\pal16l8.e6 [not fixed]

Funny thing is, these files are considered good dumps in other sets (clones or parents), but not in these particular sets. I wish there was a way for ClrMAMEPro not see them as unneded and automatically keep them, since they are (most likely) good dumps. Maybe just marked wrong in the source?

Then I launched scanner 0.02.1 to see how it would react to the new, compact collection created by rebuilder 0.09. It warned me about misplaced and wrongly named CHD disks, so I let it fix these issues. Right now, it's stuck at 48% for almost 10 minutes. I find it strange that the number of folders containing CHDs is still the same (947 from the split collection).

[Roman]

Hi, first of all, thanks for the donation.

Long thread, so I hope I can answer everything here...

> "I like the idea of a full merged ROMset, but ClrMAMEPro 4.048d can't do it the way I want, with clone subfolders"

You can do this. It automatically does it when you got merge collisions (same filename, different hashes inside a parent/clone relationship). If you want subfolders everywhere, you need to enable settings->full merge mode -> hash collision mode

>"replacing commas by periods in some machines' refresh rates"

well, yes and no. the tools do some changes to things which would lead to a bad filename (e.g. "." can't be at the end of a filename in Windows or |,*,? can't be within a filename, that's why they get replaced for machine/rom/disk/sample names). The refresh rates , . replacements are however done by pugixml which does some localized transformation of decimal numbers automatically. No big deal though.

> "After MAME finished verifying both collections, I had 47,016 txt files on each drive. 5,323 of them had 0 bytes"

Erm...does MAME creates them? Because cmpro / new tools don't create dummy sets

> "kizuna4p is a clone of kizuna, but kizuna4p.zip isn't created by ClrMAMEPro"

old cmpro has an option for "fake or dummy clones". Profiler->Options->Parsing Options->Don't create dummy clones.
new tools would handle them as sets on their own


>"Side note: I am puzzled as to why some machines which don't have ROMs receive OK"

Well, guess you need to ask MAME devs why MAME's audit does so. Yes, one is a "non runnable" device. While the other is a machine also representing a softwarelist holder.

>"Continuing, the new rebuilder inserted 13 files marked as NO GOOD DUMP KNOWN inside 11 zipfiles"

I can't repeat this. Can you tell me your rebuilder settings? Can you repeat it? If yes, can you send me one of the input folder archives which lead to this?

>"It warned me about misplaced and wrongly named CHD disks".

Would be interesting to know which ones. If the rebuilder created them, they should be in the correct place. Can you give me information about which files are affected and what the message is? Maybe run a non-fix scan and send me the scan result (either from "scans" folder or a copy to clipboard via context menu)