On 03/05/2010 06:09 PM, [email protected] wrote:
On Fri, Mar 05, 2010 at 05:46:38PM +0100, Milan Jurik wrote:
For now after elimination of file system caches and download
deliverables IPS is (at least) 3x slower than YUM/RPM. I am not IPS
internals expert nor others involved on bug submittion so we have no
good possibility to discover the reasons for this performance issue.
If you don't understand how our system works, it's going to be difficult
for you to make any rational decision about whether our approach is
correct or not. This leads me to my other point about this bug, it's
obvious that the participants are not behaving rationally. The bug in
question has quickly become a thread for reasons why the submitters
don't like IPS, instead of any constructive discussion.
In fact, I have happened to stumble upon this and file the bug (after
using couple sustaining engineers as sounding board) to make sure
appropriate attention is drawn to the issue.
There was an immediate effort to try to blur the focus by bringing up
unrelated feature lists and packaging systems. Speaking of behaving
rationally. :)
The whole point is. It *is* slow, and IMO (and apparently I am not the
only one), beyond point of reasonable trade-off for coolest feature set
ever. Feature rich might be good, if basic functionality is not hindered.
Also I do not believe customers generally try to understand the system
to have good knowledge of why is it so darn slow. That is coder's
problem. They just expect it to perform at least OK, not really
interested in explanations.
The point that Shawn was trying to make, and that the other participants
in this bug seem to have missed, is that performance is still a work in
progress. The bug database is for specific problems with a cause and a
solution. It's not a forum for rants or general nonsense.
The point that was repeatedly made. It's really cool packaging. So cool
that performance does not really matter. Accompanied by, if you do not
do the profiling for us (and I presume give us code to fix this), do not
submit a bug and complain elsewhere. But we are not interested in your
expectations (or what you envision customer expectations) of packaging
system performance.
If you're able to identify a specific performance problem, we're happy
to look into it. However, filing general complaints isn't useful.
Moreover, the comparisions that were made in the bug report are vague
and hard to interpret. Much of this is like goint to the truck
dealership to complain that your truck doesn't accelerate like your
motorcycle.
Ran this through translate: If you are willing to do our job better, we
are willing to consider taking your contribution.
Not to mention. I myself am not a coder, wonder if down the road we hope
this gets wider adoption, we tell customers. If you are not happy,
provide us with fix, or just blog about it, but do not bother us with
generic requests like it's slow.
I was merely trying to share my finding with people who could do
something about it if they cared, to avoid some had scratching later on.
Ondrej
-j
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
