Feature or bug? (strange?) yum behaviour

[seth vidal]

On Tue, 2008-04-01 at 16:31 -0400, David Mansfield wrote:
> On F8:
> 
> [root at localhost ~]# time rpm -q gimp
> gimp-2.4.4-1.fc8
> 
> real    0m0.102s
> user    0m0.032s
> sys     0m0.008s
> 
> [root at localhost ~]# time yum -C list installed gimp
> Installed Packages
> gimp.i386                                2:2.4.4-1.fc8
> installed       
> 
> real    0m3.450s
> user    0m2.829s
> sys     0m0.394s
> 
> As you can see, it's about the same on my machine.

on rawhide:

time rpm -q gimp
gimp-2.4.5-1.fc9.i386

real	0m0.222s
user	0m0.093s
sys	0m0.024s

time yum -C list installed gimp
Installed Packages
gimp.i386                                2:2.4.5-1.fc9
installed       

real	0m1.835s
user	0m1.468s
sys	0m0.181s


Remember, yum isn't just executing a one call, it is doing A LOT more
than rpm does.

However, doing it in 1.8s seems pretty reasonable to me.

And that's a pentium M speedstepped to 600mhz.

here, I'll write a script that does just what rpm -q does using the yum
modules:

import yum
my = yum.YumBase()
for p in my.rpmdb.searchNevra(name='gimp'):
    print p



time python foo.py 
2:gimp-2.4.5-1.fc9.i386

real	0m0.711s
user	0m0.533s
sys	0m0.046s

Of course that's not parsing  any arguments, checking for any excludes
or any other config a user may have set.

Here I'll do it direct to the rpm python bindings:

import rpm
ts = rpm.TransactionSet()
mi = ts.dbMatch('name', 'gimp')
for h in mi:
    print '%s-%s-%s.%s' % (h['name'], h['version'], h['release'],
h['arch'])
    
                
time python foo.py
gimp-2.4.5-1.fc9.i386

real	0m0.358s
user	0m0.136s
sys	0m0.024s


Again - no parsing, no handling of all the oddball stuff that rpm spits
back at you, and no integration with the rest of the interfaces we have
to deal with.

-sv