Dave Gutteridge dave@tokyocomedy.com wrote:
Yes, but I'm not sure that analogy really represents the situation I'm speaking of with Linux. Items designed in the past may not work with current technologies. That's not a hard concept to grasp, the same way I don't expect my CD player to play casette tapes.
Whoa! Now you're telling me you don't know the first thing about Windows.
Windows 95, 98 and ME were actually an 8-track with a cassette and crippled CD attached. They still use the Windows version 3/4 mode that is known as 386Enhanced. That means the MS-DOS 7 kernel was constantly shunting the CPU between Real86 (the 8-track) and a hacked Protected386 (cassette/CD) modes.
Windows NT 3/4, 5.0 (2000) and 5.1 (XP/2003) were/are all Protected386 kernels and APIs like a CD with a cassette attached. In fact, Windows NT 3.1, 3.50, 3.51, 4.0 and 5.0 (2000) are _great_examples_ of Microsoft _not_ maintaining compatibility with the Windows 3.x, 4.x (95/98/Me) versions of the same period. It was like Microsoft literally had _competiting_ cassette standards (kinda like Beta = NT, VHS = DOS).
I.e., do you know the a _majority_ of Microsoft own applications did _not_ run correctly on those Windows NT versions? That's before even looking at 3rd party appplications.
I'm not talking about diffeences in release times.
I'm not either. Windows NT 3.1 and 3.50 _predate_ MS-DOS 7.0 / Windows 4.0 (Windows 95). Windows NT 4.0 "Cairo" was _total_vaporware_ when it came out in 1997 -- nothing as promised.
I'm not surprised, nor bothered, that perhaps some software written for Linux kernel 2.4 doesn't work on 2.6.
Pretty much _all_ software that runs on Linux 2.0+ runs on 2.2, 2.4 and 2.6. The _problem_ is at the compiler/library level -- GCC and GLibC to start. Then add in specific library versions. GCC was especially nasty before GCC 2.96/3, when Cygnus (now Red Hat) finally got rid of most of the non-ANSI C++ compliance.
The good news is that you often have the source code so you can rebuild. Basically anything written for GCC 3 / GLibC 2 ports very well. That's circa 1997 (GLibC 2) and 2000 (GCC 3) on-ward, as long as you have the required, additional libraries.
But assuming two different distros have the 2.6 kernel, then why shouldn't they both be capable of running the same software?
Why does Microsoft Visual Studio break code between versions? Why can't I run a vertical application that runs MS Access 97 when I have installed MS Access 2000? Why can't two versions of MS Office co-exist?
Windows is actually _worse_ in this regard.
I must admit that partly I'm questioning this because I'm a little annoyed. The first Linux distro I tried was Fedora, and only afterwards was it clearly explained that it's a sort of "permanent beta", where stability was not guarunteed.
It's was never guaranteed with Red Hat Linux either.
I'm sorry, but I read the Fedora web site carefully, and it does not explain clearly what it is.
That's because you are expecting a "product." Fedora is a "project." There are some legal reasons for that.
I thought it was a reasonable candidate for consumer use.
Some of us use it to, gasp, build America's military might. ;->
But then someone recomended CentOS, because it's more stable.
I liken to the term "more mature."
No one said "... but it's really designed more for being a server.". Nothing was said along those lines.
It's not designed for just a server. RHEL is not. But if you want to run the latest apps, that's not what it's designed for.
Now, after spending weeks getting things like Japanese support, my Palm Pilot to work, Gnome configured, and many other trials and errors, *now*, when I want to get a DVD writing program, people are saying "Oh, well, really CentOS is not really all that good for those kinds of purposes". Where was this advice before?
*IGNORE* them. They are distro pissing on the CentOS list. This list is for CentOS, and related compatibility (e.g., I'll occassionally post equivalent Fedora RPMs which map well to a CentOS release, like FC3 to CentOS4).
In fact, I'm looking at the CentOS web site now, and in it's "Goals" section it says, among other things:
- easy maintenance
- friendly environment for users and package maintainers
Noticibly lacking is anything saying "a server oriented OS", or "not really intended to run consumer level software". Where was I supposed to come to understand that CentOS was not only a "stable enterprise class OS" but also limited in exactly how many applications it would be able to accomodate?
Did you _ever_ run Windows NT? ;-> Especially back from 1993 on-ward, _before_ Windows 2000 -- let alone even Windows XP.
Don't assume anything. You're making too many assumptions based on only _limited_ Windows exposure. Windows NT ~ CentOS (not really a good analogy/equivalent, but I'll make it).
Microsoft has traded consumer compatibility for stability in the past. Red Hat has done the same with RHL/FC v. RHEL. Microsoft no longer does so because they _hacked_ Windows NT and _killed_ everything good about it (long story).
So I'm sorry if I'm sounding like a whiner at this point, but if I have to change to another distro and again go through all the growing pains of learning how to use it as well I think I might run back to Windows world.
Your view of the Windows world is rather limited.
I mean, I've come to really like Linux for a lot of reasons, but I'm getting a little tired of the "this Linux for that, that Linux for this" confusion
Again, *IGNORE* them.
that only hardened Linux gurus can sort out.
There is a learning curve. We can't help you with that. There is a general learning curve between UNIX and Windows.
Heck, there is a learning curve between consumer and enterprise Windows too.
And remember, not everyone does DVD-Video under Linux. It's sort of "unlicensed" under Linux.