duplicate downloads - arrrghh

[Mikus Grinbergs]

Running 'smart --gui' 0.42 on an x86-64 system with SuSE 10.1.

As far as I can remember, I have not manually specified 2nd party
(SuSE) channels to this system.  I have installed all four smart-
related packages -- they have "automagically" set up such channels
for me.  I do __not__ know whether there exists any overlap --
such that more than one channel would point to repositories
which were different, but provided the same packages.  (The
"automagically" defined channels *do* specify multiple mirrors.)

I recently ran a "look ma, no hands" smart upgrade, which fetched
and installed a new kernel.  What I do following any 'update' is
View -> Hide old, and see if anything interesting has shown up.
I noticed the kernel-source package there, so for it I clicked on
Install, and subsequently on 'Apply'.  Then I went out of the room.

On coming back, noticed that smart was downloading (simultaneously)
__TWO__ kernel source packages.  Both had the *same* package name.
Unfortunately, smart on its 'Downloading' panel "compacts" the URL
of the files it is fetching, so I cannot tell which servers these
duplicate packages were coming from.


I happen to be on a *slow* dial-up line.  As you may imagine,
having smart download the same kernel source package twice means
that my effective communication-line throughput is halved.  In
this case, it meant that it took about two hours longer for this
particular "transaction as performed by smart" to finish !!

PLEASE, do not download the same package TWICE.  Preserving my
communication-line throughput is more important to me than
being able to continue if the primary server is interrupted.

Also, please do not "compact" the URL on the files shown on the
'Download' panel.  I'd like to see *where* something is being
downloaded from.

mikus


p.s.  Smart completed what it was doing once one of the
      downloads was finished.  (I believe the other download
      still had some megabytes to go.)  I looked at kernel-
      source packages with Tree style -> Channels, and there
      was only *one* channel (SuSE Updates) identified as
      providing that package.  So I believe that the duplicate
      download occurred using a mirror in the *same* channel.