Cult of personality

Re:"Core Team" models need to die. (Score:4, Insightful)
by stienman (51024) <adavis&ubasics,com> on Wednesday December 31, @09:58AM (#7844971)
(http://www.ubasics.com/adam | Last Journal: Wednesday August 06, @01:01PM)
"Core Team" Development models are out-dated and sound more M$'ish than Open Source'ish.

While several projects continue to use the "Core Team" model, like FreeBSD, in my opinion, the politics involved ain't worth it.

Uh, say again? Are you saying that open source software favors one political structure over another?

So if a core team is bad, what about Linux with essentially a technical dictator^H^H^H^H^H^H^H^Hdirector? And do you believe the MS uses a core team to direct development of their software? They have a simple hierarchy, like most succesful businesses.

While it may be warm and fuzzy to say that open source == no core team, the simple fact is that different political structures are good for different projects during different phases of their life. Linux has gotten too large to be developed by a single developer, so Linus has changed the political structure to fit his needs.

Furthermore, this doesn't mean the end of the core team for XFree, only the end of a core team. They haven't spelled out a change in structure, only a change in personel.

-Adam

 

Yesterday's war heros == today's dictators (Score:2)
by iamacat (583406) on Wednesday December 31, @12:05PM (#7846056)
(http://u1.netgate.net/~snowcat/)
Nobody doubts that Linus did a lot of good to the world and still does to this day. But I do think he has a particular vision for Linux and his reputation keeps people who have some disagreement from forking or working on another OS.

Say, I want to contribute a USB kernel module for my webcam. So I read a nice doxygen reference about the virtual methods I need to override in LinVideoCapture class. Then I write my driver and put binary and optionally source on my website. I shouldn't have to update/recompile it for a couple of years, until the next major rewrite of the kernel that changes LinVideoCapture signatures.

Umm... Actually it's more like I need to read lots of uncommented C code for other drivers that access kernel data structure that seem to have nothing to do with video or USB. Then when I am done I need to make source code changes every couple of month and provide a separate binary for non-SMP, SMP and so on kernels Redhat ships.

I wrote many TSRs/drivers in ASM for DOS and it's a joy compared to ASM-like C for Linux. Yes, OSes are more complicated now. But they should also provide higher-level interfaces to compensate.

Or I can just run Windows (well, MacOSX in my case) and spend the same effort to write a nice Java video-editing app. I would love to have an OS that people really use and that makes contributing to the kernel pleasent though. Like a forked Linux distribution with kernel interfaces, libc, ld.so, threading model, gcc C++ ABI... thought though once and stabalized for long time.