Re: More benchmark bullshit, and Linux luser mating calls... (was Re: Linux betas NT in TPC testing, running Oracle8

On 26 Apr 1999 15:09:00 GMT, The Ghost In The Machine <ewill_at_lexi.athghost7038suus.net> wrote:
>On Fri, 23 Apr 1999 16:54:19 -0400,
>Joshua Schaeffer <electric_ninja_at_email.msn.com> wrote:
>>>Oh, linux has _plenty_ of flaws, just fewer than WindowsNT for example.
>>
>>Would you care to elaborate? And be specific.
>
>I can at least state that Linux has a nasty flaw -- but it's mostly
>an internal one, and may not have much applicability (it is a big
>C program, after all).
>
>It uses pointers in its structures. Naughty naughty! :-)
>
>(I've been corrupted by a famous person who gives out C++ lectures around
>here whose name completely escapes me at the moment (grr) --
>"Pointers are bad". Mind you, it's not clear whether we want
>to have C++ and STL [*] in the Linux kernel... :-) )

If he is doing lectures that say "Pointers are bad," then I would posit the theory that he's an idiot.

Computers are doing "pointer stuff" all the time. Continually.

That's what addressing is all about, and if you take out pointers, you remove what computers do all the time, and make them unable to address either data or programs.

It is quite a different thing to claim that "Writing programs that treat pointers as first class objects is a bad idea."

And that is actually *still* a questionable claim; even if you're using a language that hides pointer references (as do languages like Lisp or Java), at the system level, *somebody* has to write some base code that does the "pointer stuff" in order to translate references into logical and physical addressing.

It is entirely reasonable to expect to *need* to do this sort of thing within an OS kernel like Linux. You may be able to get away with hiding pointer references in applications running atop an OS, but not in the writing of the OS.

>If this is the worst flaw of Linux, then I for one am happy with it.

The "worst flaw" of an operating system is that it has to get down to doing low level stuff like direct memory addressing and direct access to physical devices. Of course, that's the whole PURPOSE of an OS, which establishes that this isn't relevant as a flaw.

>I suspect there are performance issues in ext2fs as well, but it's
>not clear that I'd want to go to NTFS, a journaled file system, or
>something like HPFS (Apple's system) just because of that.

There are two kinds of people:
1. Those that know enough to know how to implement filesystems, and 2. Those that don't.

Those that don't know how to implement filesystems are reasonably likely to lack the knowledge necessary to evaluate the *CLEAR* superiority of one FS over another.

-- 
"The main reason for open-source gets developed is need.  If the need
is there, then the software will get written.  If it isn't really
needed, then the lack of software is hardly a problem, right?" --
Almost Brian Hurt's Words <bhurt_at_visi.com>
cbbrowne_at_hex.net- <http://www.hex.net/~cbbrowne/lsf.html>