Phoronix: Ryan Gordon Is Fed Up, FatELF Is Likely Dead
The news just keeps rolling in today... Besides VIA trying again to submit their kernel DRM, learning about KDE 4.4 features, announcing AMD's UVD2-based XvBA finally does something on Linux, the release of the Linux 2.6.32-rc6 kernel, and GNOME 3.0 likely being delayed to next September, we also have news this evening from the well-known Linux game porter Ryan Gordon (a.k.a...
I'm really sorry to see him get burnt like that, but that was to be expected. This is a major undertaking. He created something that changes the fundamentals of Linux, spanning many projects. It's very difficult to get consensus for that. It really has to be absolutely clear that this is a necessity before it can take off.
And I don't think it's that good an idea. It sounds very cool, a binary that works everywhere. So I like it in a gut-feeling way. But when you look at it objectively, when are you going to need this? Why not just put two binaries in the same package and install the correct one? Yes, there is a lack of a unified package manager, but that is being worked on.
I think Ryan underestimated the difficulty of getting other people to buy into your idea. Look at how many people thought about solutions to the package management problem. You've got RPM in the LSB, you've got Autopackage, you've got PackageKit, you've got PackageAPI... will one of those be the final solution? Maybe we'll know in 10 years.
.deb already supports multiple archs in a single file, doesn't it?
I don't know if it specifically supports multiple architectures, but you could always leave the architecture field blank, so it installs on any arch. But then you still have the other 50% of Linux users that don't use dpkg. So that's still a problem that needs solving. But when that is solved, Ryan's itch has been scratched too.