optparser attribute missing in cli.py (warning-message-printing fail when --part is given a nonexistent file)

I had this exact problem, and I just figured out what was going wrong.

It turned out to be that I'd passed a nonexistent file to "--part" in my vmbuilder command. (In the bug-reporter's case -- I'm not sure whether --part accepts relative paths or not, but if you're sure that vmbuilder.partition exists, you might try specifying the *full path* to that file.)

So as I understand it, this problem arises from user error in specifying the "--part" argument. HOWEVER -- there is still a bug here -- the vmbuilder script tries to be helpful with a warning message, but it ends up falling over and dying because it tries to print the message via a nonexistent attribute. So instead of a warning message, the user gets a cryptic exception backtrace.

> hypervisor.optparser.error("%s parsing --part option: %s" % (errno, strerror))

My python-fu is not strong, but from looking at the rest of the cli.py file, I think we want "self.optparser", not "hypervisor.optparser".

Testing that to see if that fixes it locally...

self.optparser didn't work -- tried that because there was one other usage of that in the file, but it gives me
>AttributeError: 'CLI' object has no attribute 'optparser
So the existing "self.optparser" line is probably just as busted as this bug's "hypervisor.optparser" line. :)

Trying just plain "optparser" now, since that's used many more times elsewhere in the file...

that didn't work -- reading the file slightly more intelligently, I see that "optparser" is a local variable initialized in "main" and passed to its helper functions. It's not passed to set_disk_layout, though (but it should be).

Here's the patch as described above. Just passes the "optparser" variable to set_disk_layout, and uses it to print the error instead of using the nonexistent "self.optparser" & "hypervisor.optparser".

I've confirmed that this works locally (i.e. it makes vmbuilder print the warning message instead of throwing an exception).

Assigning bug to me, since I've attached a fix.

I'm not sure who to ask for code-review from (if needed) or how to get the patch committed, though -- Soren, could you perhaps help with either of those?

Daniel, thanks very much for your contribution to Ubuntu.

I am unable to reproduce this bug because for some reason vm-builder just hangs for me on both Lucid and Precise. However, from visual inspection, I think your patch is spot on. Note however that it does not apply cleanly to Precise.

See lp:~barry/ubuntu/precise/vm-builder/bug-612082 for a branch that resolves the conflicts, and which I propose to upload to Precise. Once the patch lands there, we can prepare an SRU for Lucid, and I am happy to help with that if you'd like. I will also link a bug task with upstream so that they can apply the fix there too.

I've also uploaded the Precise package to my PPA, which you can enable with:

$ sudo add-apt-repostory ppa:barry/python

Merge proposal for trunk:

https://code.launchpad.net/~barry/vmbuilder/bug-612082/+merge/80615

On further reflection, I think the patch is safe to upload to Precise, and I will do this. I'm not entirely convinced the bug meets the SRU guidelines for Lucid

https://wiki.ubuntu.com/StableReleaseUpdates

so I won't prepare an SRU for it. I have however uploaded vm-builder_0.12.4-0ubuntu0.3~ppa0 to my PPA so you should be able to grab it from there and at least test it to make sure it fixes your problem. If you think an SRU is warranted, please review the guidelines above and file a separate bug for that.

Thanks again for your contribution to Ubuntu!

This bug was fixed in the package vm-builder - 0.12.4+bzr471-0ubuntu2

---------------
vm-builder (0.12.4+bzr471-0ubuntu2) precise; urgency=low

  [ Daniel Holbert ]
  * Fix warning messages when --part is given a nonexistent file.
    (LP: #612082)
 -- Barry Warsaw <email address hidden> Thu, 27 Oct 2011 15:52:56 -0400

Thanks!