Skip to Content.
Sympa Menu

sm-discuss - Re: [SM-Discuss] tmpfs

sm-discuss AT lists.ibiblio.org

Subject: Public SourceMage Discussion List

List archive

Chronological Thread  
  • From: julian AT openit.de (Julian v. Bock)
  • To: Phil/CERisE/KG6MBQ <cerise AT littlegreenmen.armory.com>
  • Cc: Source Mage Discuss <sm-discuss AT lists.ibiblio.org>
  • Subject: Re: [SM-Discuss] tmpfs
  • Date: Thu, 25 Jul 2002 11:49:31 +0200

Hi

>>>>> "P" == Phil/CERisE/KG6MBQ <cerise AT littlegreenmen.armory.com> writes:

P> I can think of a couple of work arounds. You could chroot into
P> a directory and install things to the appropriate places. You
P> could then relink things accordingly.

*complicated*

P> You could create a new
P> directory as you would with tempfs and do all sorcery related work
P> within it. You could then delete leftovers accordingly.

I would have to do that manually. It would be done automatically after
a reboot.

P> You could
P> have a temporary file mounted as a loopback and do all work in
P> there.

In what way is this better than tmpfs? Furthermore, it would be
slower since there is an extra device layer (loopback).

P> The important part here is that Sorcery as a good packaging
P> system should clean up after itself.

It should not clean up after a compilation or installation failure.
How are you supposed to debug the problem without the source
directory?

P> Tempfs is a simple idea
P> thrown at a problem.

That is why I like it. It is much simpler than any of the above.

P> I believe that the benchmarks have bourne out
P> that using the native filesystem may be a better idea.

The benchmarks don't clearly show that tmpfs is slower. Furthermore,
the differences are so minimal that you might be able to compile your
kde or gnome suite a whole minute faster.

P> I don't
P> believe that having Sorcery clean up after itself and having
P> Sorcery use the native filesystem are mutually exclusive.

Sorcery does not do the cleanup. Again, I am speaking of the
directories that are left after a compilation failed.

Of course, it could be done without tmpfs, but tmpfs is sooo simple.

Even if it is slightly slower (< 1% from what I have seen so far;
sometimes even slightly faster) there is no reason to throw it out.

Julian




Archive powered by MHonArc 2.6.24.

Top of Page