Print

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

[oxyandy]

As you can probably guess, right at this moment, I am attempting to make a Reduced CMP Folder..
Dat & ROM set to reproduce bug.
That way I can send you whole thing..
You extract Set folder:
C:\Bug\Set
C:\Bug\Fix_Files
Open CMP and you shall see, I'll need time to get it just right..

[oxyandy]

Ok, I take a huge step backwards
 
This is something I have never seen before.
Is only 4 7z archives it is happening with.

Archive 1: 3super8.7z
Test with WinRAR 5.10



Archive extracts fine with WinRAR 5.10 too.
I even checked CRC32 of extracted files, yes perfect..
(I did this with all 4 7z archives = same results)

BUT, when I open (any of the 4 7z archives) with 7z File Manager 9.28a
I get



I sent this archive to torrent7zip and it extracted & rebuilt just fine.

Crazy huh ?
Same with other 'problem' sets...

So here's a sample for you.
What makes this 'type' of 7z so unique ?
Tell me what you find ?
<Attachment Snip>

[oxyandy]

So anyway, I 'fixed' my set.. Ignore my stupid bug report 
I simply removed these '4 unique' 7z archives from my set...
Played with them a bit:
All 4 test fine with WinRAR.
All 4 extract fine with WinRAR.
All extracted files proved to match original CRCs (no corruption upon extraction)

All 4 can not be opened by 7z File Manager 9.28a, only version I have tried, wont be using others.
Latest versions of 7z File Manager complain t7z files have data past end of file.
(All same error - "Can not open as archive")
All 4 can be extracted and converted to t7z format, by t7z.exe

CMP can open & Rebuild all 4 as Source files. (No files are skipped)
However, CMP can not add to them if they exist in Rebuilder's Destination Path.

Quirky stuff 

EDIT: Oh and these 4 quirky 7z archives must have been made through CMP:
I don't have the patience to manually go adding files/archives to a set.
That's what a ROM Manager and DAT is for.

[Roman]

hmm...well...as you know cmpro calls 7z exe to add files to an archive....so...it was your 7z which created the files... ;-) maybe you used some buggy alpha in between....

...I had no problems with your attachment (7z 9.22)

[oxyandy]

...I had no problems with your attachment (7z 9.22)

Yeah, Oddi said also no problem opening this sample 7z with (7z 9.34a) either.
I'll remove it now

So I checked the CMD 32bit version of 7z.exe 9.34, same exact behaviour as 7.28a 7z CMD line exe..
(Wont add to these 4 7z archives either)
Restored CMP back to 9.28a 7z cmd exe, cause some of us have seen quirks with newer betas.

I give up, I am just happy there is nothing for you to actually do here.
I know you 'have enough on your plate' for the moment with CMP, I really didn't want to add to it.
haha

[Roman]

so actually we only have your unpacked rebuilder thingie left over....

[Roman]

ah...I found out why your source / destination setup is reuilt again and again and again, even if the files already exist...I did not cheat ;-)

It's because you use chd files as roms....and there is one place in the rebuilder code where this makes a difference and so your files are not checked for existance in the destination.....and so they get rebuilt....again...and again...

hmm...will see if I can change it....


http://mamedev.emulab.it/clrmamepro/binaries/cmp20140814.rar
(be warned...this build also includes the changes regarding possible full merge conflicts and use of subfolders)

[oxyandy]

(be warned...this build also includes the changes regarding possible full merge conflicts and use of subfolders)

Great, I'll try the sub-folder additions out right away
Then in the morning, I'll see if I can clean up miss matched CHD versions using a dat of current 'actual' files.
Knowing this time, if I have any duplicates they should be deleted (rather than backed up)
Thanks

Any news on the Model 110 thingy ?

[oxyandy]

Ok, I'll try the CHD thingy...

Cause I didn't get far, sub-folder test




CHD rebuild, works very fast & perfectly... nice fix

EDIT: I first got message about 18w clone and rename with sub-folder, saw file was going to be moved to parent.
I allowed in CMP looks like it was doing something... no file dates changed on archives.

[Roman]

Model 110 should be gone since some test versions...
your rebuilder problem gone, too nice...

crash...well...as I said....not yet fully tested ;-)  ...and I removed the package...good to hear the rebuild thing is gone


I know what the crash is all about (index error...most likely based on the change of name and resort....)...but I need some more time next week to find the (most likely simple) root issue...

so...delete the build and use the official cmpro ;-)

[oxyandy]

Well, this DAT & Files doesn't repeat crash.
But it does show archives remain the same after a run of "Scanner".
Set everything merged 7z
Untick Samples. Tick parse merge tags..
Use these 7z's as your ROM set
Then Scan away..

EDIT: I will keep using this CMP as it seems for everything else is OK
Plus a couple of recent changes I am now dependant on for speed.

[Roman]

as I said..do not use yesterdays build :-)
See it as a fixed version for your rebuilding thing..nothing more...it has known other issues

[Dullaron]

Good to know that you are working on the fixes and changes.

[Roman]

Well, this DAT & Files doesn't repeat crash.
But it does show archives remain the same after a run of "Scanner".
Set everything merged 7z
Untick Samples. Tick parse merge tags..
Use these 7z's as your ROM set
Then Scan away..

EDIT: I will keep using this CMP as it seems for everything else is OK
Plus a couple of recent changes I am now dependant on for speed.

I blame 7z... ;-) Works fine with zip.... your 7z files are solid....modifying them won't work....since 7z will simply fail on deleting single entries (to reorder them)....

[oxyandy]

I wrote a 7z Front_End.exe captures all Inputs/Outputs of CMP & 7z processing.
Some message boxes & text logs.
It also confirms that 7z cmd line 9.34a is a complete flop when using with CMP.

##########################################
7-Zip 9.34 alpha  Copyright (c) 1999-2014 Igor Pavlov  2014-06-22
Scanning

Creating archive D:\cmp4014__32\temp.7z

Compressing  qwerty

Everything is Ok

Kernel  Time =     0.000 =    0%
User    Time =     0.031 =  200%
Process Time =     0.031 =  200%    Virtual  Memory =      4 MB
Global  Time =     0.015 =  100%    Physical Memory =      5 MB


7-Zip 9.34 alpha  Copyright (c) 1999-2014 Igor Pavlov  2014-06-22

Updating archive D:\cmp4014__32\temp.7z


Everything is Ok

Kernel  Time =     0.031 =  100%
User    Time =     0.015 =   50%
Process Time =     0.046 =  150%    Virtual  Memory =      2 MB
Global  Time =     0.031 =  100%    Physical Memory =      4 MB
##########################################


While it passes CMP's internal initial test..
In the real world it fails with many archives..
So Roman if you do any tests, please use 9.28a of 7z.exe

##########################################
7-Zip 9.34 alpha  Copyright (c) 1999-2014 Igor Pavlov  2014-06-22


Error:
There is some data block after the end of the archive
Not implemented



Kernel  Time =     0.031 =   66%
User    Time =     0.015 =   33%
Process Time =     0.046 =  100%    Virtual  Memory =      3 MB
Global  Time =     0.046 =  100%    Physical Memory =      4 MB


System error:
Not implemented
##########################################

I learnt today that this 9.34a 7z.exe could not process the files in my earlier post...(Attached @ post #70 in this thread.)
9.28a works perfectly  I had forgotten I still had 9.34a in there after attempting to prevent the crash while scanning the merged MAME set.

[Roman]

so I switched to the last official (9.20) one from www.7-zip.org (I know there are other releases on sourceforge) and it works perfectly fine with your #70 post files.....most likely because the "rn" rename command cannot be used with this version...

gonna search a "rn" support version now and will check what the reason for failing could be...


ok...used 9.28...with enabled rn support...also no issues with #70 files....so I guess I can put this problem on the list "weird things with new 7z versions"

[Dullaron]

I been using the 9.28a very long time with the clrmamepro. More stable than the current.

[oxyandy]

Hi Dullaron,
yes I have been a huge advocate of 9.28a since I did lots of testing with 7z & CMP (quite a while ago)
From the output of 9.34a we see:

7-Zip 9.34 alpha  Copyright (c) 1999-2014 Igor Pavlov  2014-06-22


Error:
There is some data block after the end of the archive
Not implemented

This means 9.34a will simply refuse to do 'anything' with a 7z created with t7z..
Not ideal for many of us 
7-Zip 9.34 alpha = no, no, no !
There maybe other 7z versions (9.25a?) which are useful, but 9.28a was the one I settled on after exhaustive testing
Users will also find Newer Installations of 7z UI will give constant annoying messages about 'this type' of 7z archive too.