[conspire] $YOUR_FAVOURITE_FILESYSTEM sucks
daniel at gimpelevich.san-francisco.ca.us
Thu Sep 27 09:46:02 PDT 2007
On Wed, 05 Sep 2007 14:28:19 -0700, Rick Moen wrote:
> Quoting Daniel Gimpelevich (daniel at gimpelevich.san-francisco.ca.us):
> [JFS, and sharing of drives among distributions. Mutibooting, I
>> Sadly, this is an area where Ubuntu is a bit behind the curve. Ubuntu
>> still insists that ext3 is the best default.
> Bearing in mind that I don't multiboot, I'm still of the mind that ext3
> has, over the last few years, been the safest (of available journaling
> FS options) for data protection, in the general case. It's the best
> debugged on account of good maintenance and very wide testing, has
> mature and well-designed fsck and other ancillary utilities, tops out at
> reasonable sizes (say I, having not gotten that exabyte hard drive yet),
> and is designed defensively with the failure modes of commodity PC
> hardware in mind.
This was not a reference to multibooting, but to the observation that
there are a high number of things that can and do go wrong with ext3
filesystems, whether exposed to casual or extreme use, or anything in
between, relative to the possibilities of things going wrong with JFS. It
is very true that ext3, in theory, leaves more possibilities for recourse
after things go wrong, but I now consider the noticeably reduced need for
such possibilities to outweigh that.
> I'm delighted that you two have a high opinion of JFS, because that
> raises some hope that you'll do all that great pioneering QA work that I
> always appreciate someone _other than me_ doing.
A hope that I shared -- but I saw enough of people other than me
pioneering the use of JFS without issue that I decided, after quite some
time, to try it for myself in earnest, also without issue. I happen to be
writing this message from JFS as continued demonstration of that. The
pioneers are well past establishing their monopoly on receiving arrows in
the back for this one. The only arrow left for anybody else seems to be
the fact that a JFS filesystem not cleanly unmounted will refuse to mount
at all ever again until the journal is manually replayed. Although that
should sound like a no-brainer, the difficulty is that the mount command
will refuse to recognize that a JFS filesystem is even there in such a
>>> I find the buntu crowd so proud of their choices they lose honesty and
>>> like to describe it as "TENDENTION" both as dishonesty from excessive
>>> partisanship and the medical term for an undeveloped fetus.
>> This may actually be more of that "California" thing that Rick was talking
> I think I may have lost track of what this concerns. Bruce had made
> some mostly-unparseable and vague reference to inveighing against
> GNOME HIG provisions, right?
I think that Bruce was extending GNOME's "religious debate" HIG
justifications to cover a perceived arrogance on the part of installer
designers when it comes knowing what's best for the user in spite or
absence of the user specifying otherwise. Additionally, there was more
"Foo failed upon the introduction of Bar, so Bar caused Foo to fail."
> My referenced wording must have been that bit about "the rationalisation
> stage required following all screw-ups". It's the oddest thing, and
> just might be yet another bit of damage to the national culture that
> California must one day answer for: You point out that party A has
> messed up, citing reasonably objective evidence of same.
> Party A's response _could_ reasonably be one of the following:
> o "You're mistaken about your facts." Refutation follows.
> o "Your implication of wrongful effect is in error" (the "so what?"
> o "You are correct, but other good effects outweight that" (cited).
> However, the classic California-culture response is none of those,
> but rather an hours-long digression onto protestation of good motives.
> Which is of course completely irrelevant to the point -- and raises the
> reasonable suspicion that some arbitrary number of future screw-ups will
> be justified the same way.
It is precisely because such a digression would be irrelevant that it is
also not mutually exclusive with any of the three above responses. It is
neither a justification nor a lack of justification, but merely a waste of
time and effort. In the end, whether one considers doing something to be
wrong or not isn't all that relevant to whether or not same goes ahead and
does it anyway. That's the real issue.
More information about the conspire