Fedora 20 Beta released

Fedora 20 Beta released

Why would I want to write a program (apart from distro tooling) for some specific distribution? Doesn't make much sense, considering the fragmentation of the distro space. What's the supposed benefit over just writing to the APIs of the upstream libraries I want to use?

Most software developers want to ship binaries. Most users want to use binaries (even most FOSS die-hards use binary distributions and not Arch or Gentoo). Which means “writing to the APIs of the upstream libraries” is either not an option or poor option. Couple that with the fact that most software developers out there don't want to ever show sources and you get two alternatives: one poor and another which is even worse.

A cross-distro approach (obviously) would be of far more utility.

Only if it'll be actually usable for software developers.

Why should I prefer it over the tools I like and already know?

Hmm… That's good question. For most developers tools [they] like and already know includes exactly one item and indeed if you explain how to use them to build binaries for all (or at least few major) Linux distributions out there what these tools you may attract some developers. But so far noone managed that feat.

The actual appeal would be a list of interfaces guaranteed to be available on distro X, Y and Z, not some kind of downloadable SDK.

Why? For most developers out there Desktop Linux comes at the fifth place! It was third place, now it's fifth. First four places are Android, iOS, MacOS, and Windows (not necessarily in this order). Well, sometimes it's all about HTML5, but this approach is still hard to use for many types of programs. Linux support is always, always an afterthrough. Yet most distributions act as if they are a kings of the hill and may dictate rules. That's just crazy and irrational.

Ubuntu is the only distribution which finally, finally is playing by rules set by “big boyz” and yes, they achieved some success. There are thousands of applications in Ubuntu's store, for example. Far cry from half million of applications in other big stores, but if you compare the userbase that's quite an achievement.

We'll see how the story will go, but my personal opinion is that Linux Desktop is “on borrowed time”: Android (and ChromeOS) are not quite ready to replace developer's desktop but they are slowly moving in this direction. Situation is quite similar to first 10-15 years of IBM PC (and clones) development. IBM PC killed the workstations and it looks like Android is poised to do the same for PC. After that point there are quite real chance that traditional GNU/Linux distributions will just die off.