Negative Review of FreeBSD 5.4

[Astrodog]

On 6/2/05, Samy Al Bahra <samy at kerneled.org> wrote:
> On Thu, 2005-06-02 at 10:47 -0700, Neal E. Westfall wrote:
> > >> Anybody have any thoughts on this review?
> > >>
> > >> http://os.newsforge.com/article.pl?sid=05/05/24/2153257&tid=8

Quite frankly, the article is actually factually incorrect on a number of points

#1. Linux x86-64 Binarys are NOT supported on FreeBSD/amd64; Linux
i386 binaries are.

#2. SMP Support is not required to boot an Opteron machine. Without
SMP support, it would simply use only one processor. Because no detail
was provided, as to the Opteron machine he used, its impossible to
tell if this is an ongoing issue, or one that is resolved by the hints
documented on the mailing lists.

#3. The Author claims ULE is "Stable", implying somehow that its ready
for production. I'm not sure I've even seen that claim on -CURRENT.
ULE is a great thing, but somehow saying 5.4 is not ready for
production, but ULE is, is simply ridiculous.

There's also no real testing done, to provide a somewhat empirical
idea of how well FreeBSD 5.4 works, merely a personal experence "I had
to re-plug-in my keyboard" type thing. No claim is made whatsoever as
to how FreeBSD runs, after the installation. Perhaps this review was
simply covering the FreeBSD 5.4 install process?

As an aside to the author, the fact that FreeBSD has problems with
Realtek GigE NICs, and Sil SATA controllers isn't very suprising. Last
time I checked, the rl driver source said something along the lines
of, "These are the worst NICs in the world", and the issues with Sil
controllers are well known. If you don't check the mailing lists, and
do a little bit of work, it should be no suprise that some random
cheap hardware you throw at it may not work as you expect it to. I am
not aware of any major server provider, using Sil SATA controllers,
and Realtek NICs in their machines. Tyan uses Broadcom, and they're
pretty much the main OEM vendor of AMD Server kit.

By testing with the hardware that the author chose to use, he provided
a review of the FreeBSD installation process on mid-range
workstations, and low-end Dual Opteron boards. That being said, even
on crappy hardware, I've never found any real showstopping issues,
beyond the R3000Z thing, which is fixed in 5.4.

If someone isn't willing to really test an Operating System, with
benchmarks, a wide array of hardware, and software, they shouldn't be
writing a review of it, that simply claims its not ready for
production.

--- Harrison Grundy
(CC'd to article author)