FreeBSD 5.2-RELEASE

[SUOrangeman]

I was averaging over 3MB/s from ...

http://ftp6.us.freebsd.org/pub/FreeBSD/ISO-IMAGES-i386/5.2

-SUO

 

[n0cmonkey]

5.3 is scheduled as stable still, right?

 

[Mucman]

[mucman@silence ~]$ uname -a
FreeBSD silence 5.2-RELEASE FreeBSD 5.2-RELEASE #5: Sat Jan 10 22:06:42 PST 2004
mucman@silence:/usr/obj/usr/src/sys/SILENCE i386

I cvsuped just before the release date, but managed to get the RELEASE source! My plan was to go to RC2

I'm hoping 5.3 will be released as STABLE... but it won't be the case if they haven't completed all that they deem necessary to release it under STABLE 5-STABLE roadmap.

 

[Sunner]

I'd love to play more with FreeBSD, I just wish they'd stop all their silly little oddities

 

[Mucman]

Originally posted by: Sunner
I'd love to play more with FreeBSD, I just wish they'd stop all their silly little oddities

Like?

 

[Sunner]

Originally posted by: Mucman

Originally posted by: Sunner
I'd love to play more with FreeBSD, I just wish they'd stop all their silly little oddities

Like?

wtf is "adduser" ?
/home being a symlink to /usr/home ?

Nothing major really, just a bunch of little annoyances that eventually end up annoying the crap out of me in the end.

 

[n0cmonkey]

Originally posted by: Sunner

Originally posted by: Mucman

Originally posted by: Sunner
I'd love to play more with FreeBSD, I just wish they'd stop all their silly little oddities

Like?

wtf is "adduser" ?
/home being a symlink to /usr/home ?

Nothing major really, just a bunch of little annoyances that eventually end up annoying the crap out of me in the end.

adduser is a replacement for useradd. OpenBSD has one. FreeBSD changed theirs to a C program though, I think.

 

[Sunner]

It was a rethorical question

I know what it is, what I don't understand is why someone came up with the utterly moronic idea to replace useradd with it.
Didn't even know OpenBSD had one, cause they still have "useradd" as well, so I'll just use that and go about my business.
With FreeBSD, I was like "WTF, useradd not found??? PATH looks ok...wtf...Im root...wtf...find / -name useradd...wtf wtf WTF???".

 

[Mucman]

Originally posted by: Sunner

Originally posted by: Mucman

Originally posted by: Sunner
I'd love to play more with FreeBSD, I just wish they'd stop all their silly little oddities

Like?

wtf is "adduser" ?
/home being a symlink to /usr/home ?

Nothing major really, just a bunch of little annoyances that eventually end up annoying the crap out of me in the end.

That's what's preventing you from using FreeBSD? Little extreme, don't you think? Nothing stopping you from aliasing adduser to useradd, and mounting your own /home.

btw, is useradd a unix standard? Or is it just a Linux thing?

 

[n0cmonkey]

Originally posted by: Sunner
It was a rethorical question

I know what it is, what I don't understand is why someone came up with the utterly moronic idea to replace useradd with it.
Didn't even know OpenBSD had one, cause they still have "useradd" as well, so I'll just use that and go about my business.
With FreeBSD, I was like "WTF, useradd not found??? PATH looks ok...wtf...Im root...wtf...find / -name useradd...wtf wtf WTF???".

OpenBSD's useradd will probably compile on FreeBSD.

 

[n0cmonkey]

Originally posted by: Mucman

That's what's preventing you from using FreeBSD? Little extreme, don't you think?

Isn't convenience and that general feeling of something "being right" worthwhile reasons to use one system over another? That general feeling is why I use OpenBSD.

Nothing stopping you from aliasing adduser to useradd, and mounting your own /home.

He says there isn't a useradd, so that might be tough.

btw, is useradd a unix standard? Or is it just a Linux thing?

Yes, useradd is generally the way I've added users on just abuot everything but OpenBSD.

 

[Sunner]

Isn't convenience and that general feeling of something "being right" worthwhile reasons to use one system over another? That general feeling is why I use OpenBSD.

That pretty much nails it.
OpenBSD and Solaris both give a nice warm cozy feeling
Linux does too, albeit to a lesser extent, but it has many other qualities that I like.

I guess I could compile an alien useradd on a FreeBSD box, but on the other hand, I feel like I shouldn't have to, but if I don't, it'll keep annoying me every time I have to use that klunky adduser thingy.

I guess my biggest beef is pretty much just "why?".

 

[LuckyTaxi]

why does it take so freaking long to compile programs i wanna install. SIGH

my sound doesnt work ... i love the simplicity, but there are quirks i really hate about it.
running the new suse at the moment

 

[Klosters]

Interesting article about "BSD for Linux User's" in yesterday's www.slashdot.org . FreeBSD make a lot of sense, it's just the philosophical
differences that need to be understood. Easier said than done sometimes. Right sidebar.

 

[n0cmonkey]

Originally posted by: lilcam
why does it take so freaking long to compile programs i wanna install. SIGH

my sound doesnt work ... i love the simplicity, but there are quirks i really hate about it.
running the new suse at the moment

Do programs compile faster on SuSE? What version of gcc? gcc 3.x will compile slowly, it's made to compile slowly. Fast compiling isn't on the minds of gcc developers, probably because they don't worry about slower machines.

 

[Sunner]

Originally posted by: n0cmonkey

Originally posted by: lilcam
why does it take so freaking long to compile programs i wanna install. SIGH

my sound doesnt work ... i love the simplicity, but there are quirks i really hate about it.
running the new suse at the moment

Do programs compile faster on SuSE? What version of gcc? gcc 3.x will compile slowly, it's made to compile slowly. Fast compiling isn't on the minds of gcc developers, probably because they don't worry about slower machines.

Well, they still support m68k's, VAX, and a bunch of other not so fast architectures(well, architectues that don't currently have any fast CPUs, before I get flamed by some VAX fanatic).

Seems to me like compile speed is inversely proportional to code speed.
In code speed ICC > GCC 3.x > GCC 2.9x
Compile speed, GCC 2.9x > GCC 3.x > ICC

 

[n0cmonkey]

Originally posted by: Sunner

Originally posted by: n0cmonkey

Originally posted by: lilcam
why does it take so freaking long to compile programs i wanna install. SIGH

my sound doesnt work ... i love the simplicity, but there are quirks i really hate about it.
running the new suse at the moment

Do programs compile faster on SuSE? What version of gcc? gcc 3.x will compile slowly, it's made to compile slowly. Fast compiling isn't on the minds of gcc developers, probably because they don't worry about slower machines.

Well, they still support m68k's, VAX, and a bunch of other not so fast architectures(well, architectues that don't currently have any fast CPUs, before I get flamed by some VAX fanatic).

Are you sure? I heard they dropped support for some architectures (as in no one bothered to continue to update the code). m68k, m88k, or vax is one of them. Can't remember for sure.

Seems to me like compile speed is inversely proportional to code speed.
In code speed ICC > GCC 3.x > GCC 2.9x
Compile speed, GCC 2.9x > GCC 3.x > ICC

That's because ICC and GCC 3.x are adding too many optimizations, and not caring about compile time. The gcc developers say they are going to work ont his in gcc 3.4, but unfortunately, it hasn't been released yet (leaving us with slower machines hanging).

 

[Sunner]

Had a quick peak at the GCC page, looks like m88k might be it, didn't find it there.
Of course I didn't exactly look very carefully either.

As for optimizations, I actually think the developers are doing the right thing, hanging on to the really old stuff out there(like pre sun4u) is kinda pointless.
I don't see what kind of recent software those kinds of boxes would be running anyways.

It's like people asking about what to do with an old 486, granted, it CAN be used as a router/firewall, or even a workstation, but really, throw the old junk away and spend $10 on a P2 or something.

I can get both a SPARC Station 5 and a 1+ for free, but I got my Ulta1 for $100, and that included a 20" Sun trinitron, so I fail to see the point in getting those old pieces of junk.

 

[n0cmonkey]

Originally posted by: Sunner
Had a quick peak at the GCC page, looks like m88k might be it, didn't find it there.
Of course I didn't exactly look very carefully either.

As for optimizations, I actually think the developers are doing the right thing, hanging on to the really old stuff out there(like pre sun4u) is kinda pointless.
I don't see what kind of recent software those kinds of boxes would be running anyways.

It's like people asking about what to do with an old 486, granted, it CAN be used as a router/firewall, or even a workstation, but really, throw the old junk away and spend $10 on a P2 or something.

I can get both a SPARC Station 5 and a 1+ for free, but I got my Ulta1 for $100, and that included a 20" Sun trinitron, so I fail to see the point in getting those old pieces of junk.

Highly portable code is often clean code. It's a classic. Cheap play machines. SHADOW IDS got started on old SPARC hardware because the Navy didn't want to spend money on hardware. x86 doesn't support W^X as well as SPARC. I want to use it as more than just a door stop. I can fit atleast 9gB hard drives (might be 18...) in my sparcstation 10 (I'll be playing with this later), so it would be an essentially free (since I have the hardware already) file server for stuff I don't access often.

 

[Barnaby W. Füi]

Originally posted by: n0cmonkey

Highly portable code is often clean code.

It can also be really nasty - to be portable, you usually have to make a lot of exceptions and special cases in your code.

And gcc 3.4 will have precompiled headers, which can possibly speed things up by a huge amount, though I think many projects won't benefit much without a slight retuning of things.

 

[drag]

In the real world portable code wins over optimized code.

It's much better to simply recompile and modify code a bit to suite a different archatecture then to have to rewrite everything.

two examples:

Unix vs virtually every other major peice of software or OS from the late 70's early 80's.

Suse Linux professional AMD64 version vs Windows XP AMD64.


I'd bet that it took a fraction of the time and effort to port Suse professional over to 64bit (even when you take all the non-Suse directly related work that went into porting the kernel and other subsystems). This is because people generally default to portability and generic code when writing OS applications. Plus the *nix design empisises portablity in it's applications. Even at a cost of ease of use and speed at times.

Were as the original windows stuff was never designed to run on anything other then x86. Developers didn't care about having to do stupid tricks to get things working or heavily optimized code, because if it made it just a bit faster or solved a usability issue then it's ok. What works works.

Now you have parts of XP that are 10-15 years old, or are at least based on modified programs created 5-10 years ago. Who knows what sort of things people did back then to make things work. How many developers that worked on systems 7 years ago are still incharge of the same parts of the OS that they were years ago. How many died, quit, or simply forgot the hacks and tweaks they made happen.

How many layers on layers of programming exist between windows95/NT 3.0 and what we have today?

Linux, Unix, and BSD still have the same issues with legacy code, but since everything was designed portable from the ground up a specific developer doesn't realy need to know exactly what is going on because a simple recompile is all you realy need in 70%-80% of the time.


Or look at other propriatory software. Look at what big messes got made of AIX, SGI's Unix, or other System V code. Once they started optimizing them heavily to suite specific hardware then they doomed those OS's to obscurity and limited lifespan.

Once a hardware platform becomes obsolete then so does all that developement and hardwork. Only one that has been ported was Sun's Solaris Unix from Sparc to x86 and that's probably because it's heavily based on BSD.

What about other OSes that were considured superior to Unix like VMS? VMS started on Vax was ported to Alpha and died when Alpha died. Marketing wasn't the only issue.

Unless of course you consider Windows NT the modern VMS-based OS, which it can be considured in a lot of ways. Thru the kernel design and peopel like Dave Culter.

(proof) (tonge in cheek, BTW)

After all, were did the name NT come from?

abcdefghijklMNopqrSTuVWxyz.

V -----> W(in)
M ------> N
S -------> T

 

[Barnaby W. Füi]

I didn't say anything about performance, I said that the code can be ugly.

 

[drag]

Originally posted by: BingBongWongFooey
I didn't say anything about performance, I said that the code can be ugly.

I know. Generally though highly optimized code is better performing, but often harder to port. While very generic code is easy to port, but often performs poorly when compared to what you can do.

That's all. Usually the prospect of having code that is usefull 5-10 years from now outways the need for speed. And probably clean code.

Or somethign like that. Sorry if I went of on a tangient.

 

[Sunner]

And I didn't say they shouldn't make the code portable, I just don't think it's worthwhile effort to spend time optimizing for old architectures that will realistically not be able to run new software anyway.

I save all kinds of old junk, often for no other reason than "I might get a use for it some time", but I do throw away the really old junk that I really can't see myself using, or the stuff that can be cheaply replaced by newer stuff.

For example, and old P90 with 16 MB of RAM and 1 GB HD, sure that could make a fine firewall, but really, I saw a pile of ~10 HP Vectra's with P2-233's in them, and probably 64 MB ram in them, laying a dumpster at work.
Same thing with really old Sun stuff, I wouldn't considder anything older than sun4u worthwhile.

 

[n0cmonkey]

Originally posted by: Sunner
And I didn't say they shouldn't make the code portable, I just don't think it's worthwhile effort to spend time optimizing for old architectures that will realistically not be able to run new software anyway.

I save all kinds of old junk, often for no other reason than "I might get a use for it some time", but I do throw away the really old junk that I really can't see myself using, or the stuff that can be cheaply replaced by newer stuff.

For example, and old P90 with 16 MB of RAM and 1 GB HD, sure that could make a fine firewall, but really, I saw a pile of ~10 HP Vectra's with P2-233's in them, and probably 64 MB ram in them, laying a dumpster at work.
Same thing with really old Sun stuff, I wouldn't considder anything older than sun4u worthwhile.

I didn't say optimize for older archs, just continue to support. And the older archs are still running fine, even if they are slow