Boot speedup with readahead

[Behdad Esfahbod]

James Antill wrote:
> On Mon, 2008-09-15 at 13:02 -0400, Behdad Esfahbod wrote:
> 
>> The question is: if one could throw RPM away and design a new one, could one
>> do significantly better?
> 
>  Of course one could, the relevant question is _how long_ would it take.
> And could it be done faster by just fixing rpm, and I've not seen any
> compelling arguments that it would be faster to throw away what we have.

I made it very clear in my mail that I'm not suggesting that we should throw
RPM away.  The question I asked is supposed to give one an idea of how much of
the problem is caused by the current implementation as opposed to the inherent
complexity of the problem.

>> Lets look back at the problem at hand: we all agree that custom-installed
>> glob-matched post-transaction triggers are useful things.  I think I can also
>> say that we agree that it should be in the lowest-level package management
>> system.  What has been up to debate so far is whether that lowest-level is
>> RPM, or that RPM is a lost case and yum is considered the lowest-level.
> 
>  That is a severe mis-reading of the discussion, the question is given
> that rpm+yum are currently how all Fedora users manage their system. Do
> we want to still require that all packaging problems should be solved at
> the rpm layer, or should we try to move up and allow some more of the
> problems to be solved at the yum layer.
>  There are many advantages to doing this, including just plain ease of
> implementation. The only real disadvantage is that apt/smart/zypp/etc.
> will become even more of a second class citizen in Fedora than they
> already are (although I'm confident that the change proposed by Seth
> could easily be ported to work in all of the above).

What exactly in my paragraph you quotes is "a severe mis-reading of the
discussion"?  You say the exact same thing as I did except that instead of
calling RPM a lost case as I did, you call this approach making
apt/smart/zypp/etc. even more of a second class citizen.  End result is the
same pig: use yum or be screwed.

You and Seth seem to be on the camp that thinks it's no big deal if one uses
RPM directly and gets into a suboptimal state.  Some of us think that that's
unacceptable.  We disagree and that's fine.  Whoever ends up doing the work
decides what to do.  I fully understand that and don't mean to imply what
people should be working on.  Hope I've made myself clear this time.


behdad