a new way to hang 7.0
韓家標 Bill Hacker
askbill at conducive.net
Tue Jan 8 14:14:40 PST 2008
Robert Watson wrote:
> On Tue, 8 Jan 2008, Matthew Dillon wrote:
>> :Matthew Dillon <dillon at apollo.backplane.com> writes:
>> :> [snip]
>> :The arrogance you display in your response is simply staggering. I
>> :should have known better than to engage in any kind of conversation with
>> The arrogance you display in your posting is simply staggering. I
>> should have known better then to reply to it.
>> What the hell kind of answer did you expect, DES?
> I think this thread fell below a reasonable usefulness threshold several
> posts ago, and I suggest the participants resolve their further
> differences via private e-mail rather than on the public mailing lists.
> Robert N M Watson
> Computer Laboratory
> University of Cambridge
Unfortunately, I'm afraid it is FreeBSD itself that
'...fell below a reasonable usefulness threshold....'.
Sometime around 5.X.
Call 6.x a 'dead cat bounce', 7.X 'masturbatory' at best, and 8-CURRENT?
'hoping and praying..' Maybe.
To wit (check the list archives. All of the lists):
- Fatal scheduling and interrupt handling issues.
- Fatal storage, I/O, and fs corruption problems with far too many controllers.
Not just ZFS.
- Fatal stack/network problems
- even .. gimme a break! *lpt* issues?
Far too many of them not on new silicon - but *on hardware that used to JFW*!
To some extent *lack* of the very sort of strong opinions that Matt and
Dag-Erling at least had the balls to express.
And that at least one is turning into fs code. Good, bad, or sideways code
remains to be seen. But putting his personal assets on the line to make the
attempt - not just his keyboard.
I don't know - and don't care - if Matt, or DES, or PHK, or PJD, or ..
whomever... is or is not - seen to be difficult, unpleasant, or obstreperous.
They've each worked hard to earn the right to their viewpoints. We can still
learn from them all.
Perhaps one *has to be* a bit rude to create a sound project in a volunteer
environment. FreeBSD might have benefited from more 'sturm und drang' these past
several years - not less.
Sort the goals before coding - not after. Nobody has to die if they lose an
argument over 'direction'. But we all suffer if the issues are not given a
serious airing up-front.
Patton's 'Good plan violently implemented right now beats a better plan delayed
a week!' applies to many things. An operating system is not among them.
FreeBSD has simply become 'too much nice', 'too much knee-jerk', and 'too little
rigorous' since 4.X days to deliver its old quality.
The once-legendary stability is long gone, and a return to that seems
subordinate to getting *something* out the door with a higher release number.
The stability under load 'niche' was *all we really had*!
'Features', 'speed', 'faster than Linux' - performance in general - aren't worth
a rat's ass if the box falls over unexpectedly and only one to three developers
out of several hundred knows how to analyze the 'why' of - and others not at all.
lpd broken? And we know it? And know how to fix it? And have not?
Unfortunately, just the tip of a very large iceberg...
JM2CW (and Hong Kong cents at that ...)
But dissent is to be managed and learned from.
Not swept under the carpet.
More information about the freebsd-current