Re: Bug#264241: Some corrections regardings statements in this bug
On Tue, Aug 17, 2004 at 12:31:55AM +0200, Florent Rougon wrote:
> [ Mail-Followup-To set to debian-tetex-maint@l.d.o.; I suggest to drop
> the bug address in followups since I think the discussion is not
> relevant to the bug report any more. ]
Agreed.
It's even the question, whether this discussion belongs to
debian-tetex-maint...
> Adrian Bunk <bunk@fs.tum.de> wrote:
>
> > The shell speed is hardly the major factor in maintainer scripts speed.
>
> I think this is mostly correct, although I have no benchmark to back it
> up.
I haven't seen benchmarks that dash was faster than bash.
But even if it is, it doesn't matter much:
The maintainer scripts of teTeX spend significantly more time with
running programs like mktexlsr or ftmutil than in the shell they are
using.
Even a factor of let's say two in the speed of the shell wouldn't make
much difference for the total time.
And if you want to speed up package installation and removal, e.g.
improving how dpkg reads it's database would definitely be of more
effect than the effect of using a different /bin/sh .
> > bash is essential.
> > A Debian system without bash is therefore by definition broken.
>
> Right. They get to keep the pieces together, but as long as they are
> willing to do so, I do not mind.
There's not much use in supporting broken systems.
> > Making an essential package non-essential is basically impossible.
>
> Really? Suppose bash is not essential anymore. Well, bash-specific shell
> scripts that use "#! /bin/sh" will be found quickly. Also, packages
> calling bash correctly will stop working with a "Command not found" or a
> similar, presumably obvious, failure mode. Add the missing dependency
> and you are done. Of course, the whole transition would take some time
> to complete, but what is impossible about that?
If you do this, you have to get it 100% right.
And there are evil special cases you have to consider. An example:
Consider a potato package that uses #!/bin/bash in the postrm script,
and that was removed but not purged at potato time.
If bash is no longer essential, and the user attemts to purge this
ancient package without bash installed it will fail.
This is just one non-trivial problem that immediately comes into my
mind, and there might be others hidden.
e2fsprogs would be an easier candidate for becoming non-essential - but
even for e2fsprogs I have yet to see an experienced Debian developer who
could imagine all possible breakages caused by this to seriously
suggesting this.
> > If an embedded system is that space limited, it would use a small system
> > based on busybox instead of a Debian installation.
>
> Maybe. All I can say is, there *are* some people out there who are
> trying to get things working this way (look for posts by David Weinehall
> on -devel, IIRC).
That much effort for getting rid of 600 kB?
Wow, there are really people who don't have anything useful to do in
their spare time...
> Florent
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
Reply to: