Home Page Forums MAME MAME MAME support for writable media

The point Vas was making is that we're not yet to the level where you can start adding the glitzy enterprise features that VMWare has. Copy-on-write and that sort of thing will be important in the future, but today is not that future.

Oh fuck off. I’ve probably done more to improve the development and user experience for MAME than anyone else in the last five years. The only “thanks” I ever get from you are whinging e-mails when you can’t build with a horribly outdated toolchain.

The trouble is, you’re getting way ahead of yourself. You’re saying we need branching snapshots when MAME can’t even gracefully deal with a floppy serialisation error. I know there’s a lot of stuff MAME needs to do better. I’ve sunk a huge amount of time into building the infrastructure to actually make that possible. MAME is so big and unwieldy that any significant architectural change is a shitload of work. It inevitably results in a pile of whining from people who don’t like change and can’t look past that to see the long-term benefits. The best outcome you can get is that nothing breaks. No-one wants to do it because it’s a thankless task, and you don’t get your name in lights as the genius who emulated whatever. Hence no-one actually wants to do it, and it inevitably falls to me.

But thanks for further confirming that you’re an ungrateful whinger.

Don't break your arm patting yourself on the back.

Hopefully this thread will stick around to demonstrate yet again what an ass you are.

I expect to have a deliberately placed chunk of code added now to break my build, again.

See, I can’t win. On one side I’ve got Olivier telling me I’m not migrating MAME to pre-standard C++20 fast enough, and on the other hand I’ve got you complaining that you can’t build MAME if you refuse to update to a toolchain that isn’t over three years old. I try to strike the middle ground, and everyone’s angry at me. I’m not intentionally breaking your build, I’m making the most of more expressive language features to make life easier for contributors. It shows in the increased number of contributors, increased quality of submissions, and the almost complete absence of whole classes of issues that used to plague MAME development. Remember the dreaded late bind errors, and devices not being found? They were a huge waste of contributors’ time before C++ machine configuration, but they’ve all but disappeared now. Do you remember the pages of MCFG_ macros that you needed to write along with a device class? They’ve been completely eliminated – stuff just works without a pile of unnecessary work.

What do you have to contribute here besides a wishlist? Anyone can wish, anyone can say a feature would be nice to have, anyone say things need to be better. You haven’t said anything in this thread that we don’t already know. What are you actually going to do to move things forward?

BletchMAME sheepishly raises its hand... ;-)

Ample's pretty good too, for the limited subset it addresses (emulating Apple II and 68K Mac on modern macOS).

Al's real grumpiness is because he won't update his Mac, and Apple made some unusual design decisions that mean the compiler version and OS version are tied a lot tighter than on Windows or Linux or *BSD.

So, what kind of discussion should take place?

Here are some possibilities . . .

1. Does MAME have any sort of policy for how mutable media changes are created or stored?
2. Should MAME provide any additional features to assist users in managing their own media files?
3. Should MAME provide any support for creating media images for a particular system outside of emulation?

Also adding, that for preservation purposes, the original (as distributed) media images (and representation of media) should be preferred for cataloging and archiving.

MAME does not handle media low-level well enough to discuss higher-level features (yet). And we understand that original media, where it can be represented in digital form, is preferred. That's Firehawke's entire life right now with the Apple II software lists.

Let's talk UX then. What's missing, what should be done differently?

What I *hope* to add is the capability to edit images which have filesystems (floppy, HD, CD) and create new ones interactively. In the oric case, it could be "start with a jasmin connected but no floppy, create a new floppy image, check bootable and start program z, add zorgon.bas as z.bas, mount it, press F1" and it boots on zorgon's revenge. There are three infrastructure things missing for that, but it's on the way.

What else could we have? I'm listening to suggestions, but they should be generic enough that they make sense in Mame.

UX, I would imagine would look like an FTP client interface. Possibly a file browser for the host system and a file browser for the emulated media. Either drag and drop or have a dedicated UI element to initiate transfers. However, I would imagine this would require something like a POSIX level uniformity of file access for each emulated file system.