smart fails on rpm with empty Provides:

[Anders F Björklund]

Traceback (most recent call last):
  File "/usr/bin/smart", line 195, in <module>
    main(sys.argv[1:])
  File "/usr/bin/smart", line 168, in main
    exitcode = iface.run(opts.command, opts.argv)
  File "/usr/lib/python2.5/site-packages/smart/interface.py", line 53, in run
    result = _command.main(self._ctrl, opts)
  File "/usr/lib/python2.5/site-packages/smart/commands/nothing.py", line 36, in main
    ctrl.reloadChannels()
  File "/usr/lib/python2.5/site-packages/smart/control.py", line 378, in reloadChannels
    self._cache.load()
  File "/usr/lib/python2.5/site-packages/smart/backends/rpm/header.py", line 300, in load
    prvargs, reqargs, upgargs, cnfargs)
  File "/usr/lib/python2.5/site-packages/smart/backends/rpm/base.py", line 108, in equals
    fk([x for x in self.provides if x.name[0] != "/"]) !=
IndexError: string index out of range

http://download.fedora.redhat.com/pub/fedora/linux/releases/10/Fedora/i386/os/Packages/opal-3.4.2-1.fc10.i386.rpm
  ...
  <rpmTag name="Providename">
 <string/>
 <string>g726</string>
 <string>gsm0610</string>
 ...
  <rpmTag name="Rpmversion">
 <string>4.6.0-rc1</string>
  </rpmTag>
  ...

[Anders F Björklund (afb)]

Workaround, show warning instead of traceback

[Anders F Björklund (afb)]

Allowing this package in later caused cdistance.so to segfault when comparing against it... :-(

[William (william-redhat-bugs)]

Created attachment 324832
terminal log showing the problem

Description of problem:

When fedora.channel is added and channels are updated, a python error "IndexError: string index out of range" occurs. See attachment for log of commands leading up to this.

Version-Release number of selected component (if applicable):

smart-1.1 in Fedora f10

How reproducible:

always

Steps to Reproduce:

1. Install smart, start with one channel in /etc/smart/channels(e.g. fedora-updates)
2. smart update
   (all goes well, cache is updated)
3. add (standard version) fedora.channel to /etc/smart/channels
4. smart update
   repo data is read successfully; after cache update begins, trouble

Actual results:

smart crashes with a python "string" error

Expected results:

cache update should complete successfully

[Anders F Björklund (afb)]

segfault fixed in http://bazaar.launchpad.net/~smartpm/smart/trunk/revision/844

[William (william-redhat-bugs)]

I have confirmed that this does not occur on x86_64 (f10). Since it looks likely that this bug is triggered by data from the download site, I added in a little debug code to base.py at line 101, viz:
        try:
         if (len(self.upgrades) != len(other.upgrades) or
            len(self.conflicts) != len(other.conflicts) or
            fk(self.upgrades) != fk(other.upgrades) or
            fk(self.conflicts) != fk(other.conflicts) or
            fk([x for x in self.provides if x.name[0] != "/"]) !=
            fk([x for x in other.provides if x.name[0] != "/"])):
            return False
        except:
         print
         print "Name: ", self.name
         print "Version: ", self.version
         print "Upgrades: ", self.upgrades
         print "Conflicts:", self.conflicts
         print other.provides
         print self.provides
         raise

resulting in the following output:
Loading cache...
Updating cache... ################ ( 39%)
Name: opal
Version: 3.4.2-1.fc10@i386
Upgrades: [opal < 3.4.2-1.fc10@i386, openh323]
Conflicts: [openh323]
[lpc10, opal = 3.4.2-1.fc10@i386, h261-vic, libopal.so.3.4.2, gsm0610, g726, , opal(x86-32) = 3.4.2-1.fc10, ima_adpcm, theora, vpb, gsmamrcodec, speexcodec, h263-ffmpeg]
[lpc10, opal = 3.4.2-1.fc10@i386, h261-vic, libopal.so.3.4.2, gsm0610, g726, opal(x86-32) = 3.4.2-1.fc10, ima_adpcm, theora, vpb, gsmamrcodec, speexcodec, h263-ffmpeg]
Traceback (most recent call last):
(remainder of traceback as previously shown)

I guess this shows that the extra empty element following 'g726' is the culprit. Unfortunately, that has reached the outer limits of my knowledge, since I have absolutely no idea what all this represents, much less whether it is right or wrong :-).

Let me know if I can help in any way.

[Kelvin (kelvin-redhat-bugs)]

I can confirm that this also happens if you make a local archive of the Everything packages and repo and use that instead.

If you then rebuild a new repo using createrepo, the same thing also happens. Basically, there is something in smart that is not handling the rpm-md data.

[Axel (axel-redhat-bugs)]

(In reply to comment #1)
> print other.provides

> [lpc10, opal = 3.4.2-1.fc10@i386, h261-vic, libopal.so.3.4.2, gsm0610, g726, ,
> opal(x86-32) = 3.4.2-1.fc10, ima_adpcm, theora, vpb, gsmamrcodec, speexcodec,
> h263-ffmpeg]

> I guess this shows that the extra empty element following 'g726' is the
> culprit. Unfortunately, that has reached the outer limits of my knowledge,
> since I have absolutely no idea what all this represents, much less whether it
> is right or wrong :-).
>
> Let me know if I can help in any way.

Great debugging! This seems to be the empty Provides: bug that is submitted here:

https://bugs.launchpad.net/smart/+bug/302345

Actually the report there mentions the same package, so it's rather 100% the same bug, and the better news is that there is a fix there available as well.

New build are underway and will close this bug once they hit the update repo.

Thanks for reporting and analysis!

[Axel (axel-redhat-bugs)]

Changing to the upstream bug summary.

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.fc9 has been submitted as an update for Fedora 9.
http://admin.fedoraproject.org/updates/smart-1.1-58.fc9

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.fc8 has been submitted as an update for Fedora 8.
http://admin.fedoraproject.org/updates/smart-1.1-58.fc8

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.fc10 has been submitted as an update for Fedora 10.
http://admin.fedoraproject.org/updates/smart-1.1-58.fc10

[Anders (anders-redhat-bugs)]

The broken "opal" package was reported as Bug #473084

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.fc8 has been pushed to the Fedora 8 stable repository. If problems still persist, please make note of it in this bug report.

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.fc9 has been pushed to the Fedora 9 stable repository. If problems still persist, please make note of it in this bug report.

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.fc10 has been pushed to the Fedora 10 stable repository. If problems still persist, please make note of it in this bug report.

[William (william-redhat-bugs)]

After upgrading to smart-1.1-58.fc10, I find that the version of base.py included in the package does NOT smart-emptyprov.diff from https://bugs.launchpad.net/smart/+bug/302345, hence the problem persists.
[root@bbsuper rpm]# pwd
/usr/lib/python2.5/site-packages/smart/backends/rpm
[root@bbsuper rpm]# ls -l base.py
-rw-r--r-- 1 root root 10446 2008-09-09 04:28 base.py
[root@bbsuper rpm]# rpm -qf base.py
smart-1.1-58.fc10.i386

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.0.1.fc10 has been submitted as an update for Fedora 10.
http://admin.fedoraproject.org/updates/smart-1.1-58.0.1.fc10

[Axel (axel-redhat-bugs)]

(In reply to comment #13)
> After upgrading to smart-1.1-58.fc10, I find that the version of base.py
> included in the package does NOT smart-emptyprov.diff from
> https://bugs.launchpad.net/smart/+bug/302345, hence the problem persists.

For some reason the patch with the bugfixes was empty for F10. I resubmitted a build and update for F10.

[Axel (axel-redhat-bugs)]

*** Bug 475779 has been marked as a duplicate of this bug. ***

[Fedora (fedora-redhat-bugs)]

smart-1.1-58.0.1.fc10 has been pushed to the Fedora 10 stable repository. If problems still persist, please make note of it in this bug report.

[William (william-redhat-bugs)]

Confirmed, problem is fixed with this latest update. Thanks!

[Gustavo Niemeyer (niemeyer)]

Patch applied in trunk, with additional testing.

[Anders F Björklund (afb)]

Released, 2009-03-04