Good news, couriersud's workaround works (I need to add it for ARM builds next ;-) The next version should build unattended on PowerPC.

I can confirm that 0.146 builds flawlessy on my G4 PPC with OSX 10.5.8 without errors or exceptions.
A very good job, really, a well done release! :-) Compliments!

Yup, I've been building on my G5 pretty much every time someone changed something to make sure pies stayed away from my face on this one ;-)

And now we have a new breakage on PPC OSX. Vas Crabb, any idea on this one?

It means the code has grown too large for relative jump instructions to reach all of it. You could try building with options to produce branch islands. I think you give "-mlongcall" to the GCC when compiling (calls are slower, but code can grow bigger).

-mlongcall is the right flag according to docs, but it didn't help

I have a suggestion, just an idea, I hope it might help.
Since there are a number of drivers that require so much computation power that older PPC machines can't simply satisfy, a solution could be to exclude from compilation some drivers in order to mantain the code smaller enough to remain within the limits of the GCC compiler.
If one of the goal of MAME project is not only the preservation but also the playability it has little sense to play (on legacy systems like the PowerMacs) a game that can't go over few FPS with scattering graphics and jerky sounds.
The idea is to have a check during the compilation that say: "if the target is a Mac PPC then skip from build SEGA Model 3, Naomi, Namco PS2, Chihiro, and so on..."
I don't know what other Mac (PPC) MAME users around think about that, but in my humble opinion this could be a good way to still benefit from (usable) driver progress without have to deal with code too much heavy to be useful for playing.

That's a terrible idea, and the MAME team is not going to do that.



I'm going to stop you right there and inform you that the goal of the MAME project has never, EVER, EVER been about "playability" as well as preservation. It has ALWAYS been about preservation ONLY.

Oh, I wouldn't say flat out that we're not going to, but it certainly adds some variables to the "do I continue to push that stone up the hill" calculus. 0.146 final may simply be declared end of the line - as far as I know most other projects gave up on PPC OS X a year or more ago.

@Just Desserts: I agree for the most part, the goal of MAME project is out of question but here we are discussing about a way to (and limited to a specific MAME port: the OSX PPC build) to still permit a compilation reducing the number of driver cutting out the most hungry in term of computation power.
I think a similar thing is already done for some MAME ports like the handhelds or the Nintendo Wii so it is only a choice of the programmer that do the port for a specific platform, this can or cannot be approved by the core MAME team, but the final user of those devices is more than happy to can play at least the good oldies on its platform.

@R.Belmont: Since I am a Mac Intel user too I could easily agree (speaking for me only) in dropping the OSX PPC support but if some classic driver refinements will be done in future development (and there will be for sure) it would be a pity to cannot benefit from this because of some heavy drivers (we are AGAIN speaking only of a specific port, NOT the MAME core project) take space in building process keeping the executable from be done.

Mr. Belmont (and other few developers) have done an enormous and great done work in keeping alive the PPC support for SDLMAME and for that we will be grateful forever. The choice could be at this point if going ahead leaving behind some drivers that legacy hardware could not simply run in useful way or dropping the OSX/PPC support at all: we will miss it but life (and technology) goes on.

Anyway, thanks again for all the PPC fun! ;-)