[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: reading an empty directory after reboot is very slow

On 2015-04-27 11:04:51 -0500, David Wright wrote:
> Exactly. Saying "But this matters in a POSIX-compliant filesystem"
> is very different from "But with the current solution". To me,
> a "current solution" is an concrete implementation, not a POSIX
> specification; the latter is a target.

I really meant the concrete ext3 implementation, but in the restricted
example concerning the last entry (since at that time, the discussion
was on this point).

> In response to my tests, you wrote "You haven't demonstrated
> anything", and not "You haven't tested compliance".

No, I was right here. When I wrote "You haven't demonstrated anything",
it was on your test with "ls -lU", and I showed you that one could have
the same behavior on "ls -lU" with an implementation that follows the
spec. As I've said: "What actually needs to be done is a real test
using readdir.", which you did *later*.

> > I agree for an end user, but see the context.
> Um, I'm here in debian-user, not ext3-users@redhat.com or
> ext4@vger.kernel.org or linux-fsdevel@vger.kernel.org where I suppose
> they chat about such things.

Discussions on the implementation also matter for the end user, not
just for developers, since the implementation has an influence on the
performance, and the performance was at the start of this thread.

Vincent Lefèvre <vincent@vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Reply to: