Hi Roman,

From a previous discussion:

> 99% of scantime is pure hd access. A full MAME scan on Windows takes 7 seconds
> (with full usage of the diskcache). The bottleneck is the hd speed.
>
> WINE's MFC and hd emulation seems to be not the best/fastest.


Well, to have a comparaison, on my Wine/Ubuntu, the full scan took 150 seconds for a file (artwork.zip) on NTFS, and 126 secs on ext3. The same file is scanned on windows within (less than) 30 secs. The difference is huge.

That brings me to this question: do you plan to port clrmame on linux? I know it works very well with wine, but it works very slowly. And fastness, when scanning thousands of files, doesn't seems redundant.

What do you think? Is it a huge task?

> What do you think? Is it a huge task?

It is.

And open source is not an issue. It's still time since it would simply take long time for an outstanding person. clrmame is over 10 years old now....


Roman Scherzer

> And open source is not an issue. It's still time since it would simply take long
> time for an outstanding person. clrmame is over 10 years old now....

Is it possible to have a way to automatically translate from the language used on Windows for clrmame to one of the languages on linux (ie, c++)? How do manage the companies developing both on Windows and Linux, did they develop twice the same program, or did they make some minor adjustments at the end?

(Please bear with me, I'm not a programmer. I try to understand.)

It depends on what they develop. The backend part, if it's using ansi/iso/standard/no special libraries stuff only is rather easy to port.

Filesystem stuff has to be checked because they simply differ, not only case-sensitive names but folder structures etc.

Gui...well...as long as no portable gui system is used, you have to port/rewrite the guipart yourself.


Roman Scherzer

Last year I went searching for a better language for cross platform development. In the end I found that Python was an excellent choice. It handles all of the issues you normally have with GUI and files system differences between OS's. Not that python is the answer but it is a possibility. If you're not up to attempting a port then would you consider open sourcing the code to another team? I realize you have a lot of code and time invested but a motivated team could probably pull it off.