Re: software-properties SRU stopped due to odd bug

2021-12-13 Thread Sebastien Bacher

Le 13/12/2021 à 18:26, Brian Murray a écrit :

This type of crash has come up before and it was always my thought that
it was an older pip-installed version of distro-info.


Would it be easy to include those installed packages as part of 
ubuntu-bug reports to help understanding such cases?


Cheers,
Sebastien Bacher


--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel


Re: software-properties SRU stopped due to odd bug

2021-12-13 Thread Brian Murray
On Sun, Dec 12, 2021 at 08:36:01PM -0800, Bryce Harrington wrote:
> On Mon, Dec 13, 2021 at 03:50:10PM +1300, Robert Ancell wrote:
> > Hi all,
> > 
> > I have a software-properties SRU [1] in bionic that has phasing stopped due
> > to an error [2].
> > 
> > The stack trace is showing the following (new) code is failing:
> > 
> > di = distro_info.UbuntuDistroInfo()
> > releases = di.get_all(result="object")
> > 
> > with the error:
> > 
> > AttributeError: 'UbuntuDistroInfo' object has no attribute 'get_all'
> > 
> > This is very confusing, since the code from this (from the distro-info
> > library) is essentially:
> > 
> > class DistroInfo:
> > def get_all(self, result="codename"):
> > ...
> > 
> > class UbuntuDistroInfo(DistroInfo):
> > def __init__(self):
> > super().__init__("Ubuntu")
> > 
> > i.e. I can't see any reason why the `di` object would not have a get_all()
> > method. Running software-properties and various test scripts on a bionic VM
> > I haven't been able to reproduce any such issues.
> > 
> > Theories that have been proposed:
> > - The user is actually running an older version of disto-info that was from
> > before this method was added (in distro-info 0.15).
> >   - Seems unlikely since there are error cases in which the install was
> > done with a bionic image.
> > - The user has a locally installed version of distro-info.
> >   - Stacktraces shows the system installed version, the appropriate version
> > of the binary installed, no PYTHONPATH set.
> > - A .pyc is being used that doesn't match the source.
> >   - Since installs have been from bionic media, it seems impossible that an
> > older .pyc could be created.
> > - Other code has removed this method.
> >   - Method is being accessed immediately, no other code seen in source that
> > could do this.
> > - Memory corruption.
> >  - Too many cases for error to be random, not seeing other similar issues.
> > 
> > Does anyone have any ideas about what might be going on?
> 
> Could it be an old pip-installed version of distro-info?
> (C.f. LP: #1848829, #1874250)

This type of crash has come up before and it was always my thought that
it was an older pip-installed version of distro-info. For quite some
time the pypi version of distro-info was out of date and missing this
option.

--
Brian Murray

-- 
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel


Re: software-properties SRU stopped due to odd bug

2021-12-12 Thread Bryce Harrington
On Mon, Dec 13, 2021 at 03:50:10PM +1300, Robert Ancell wrote:
> Hi all,
> 
> I have a software-properties SRU [1] in bionic that has phasing stopped due
> to an error [2].
> 
> The stack trace is showing the following (new) code is failing:
> 
> di = distro_info.UbuntuDistroInfo()
> releases = di.get_all(result="object")
> 
> with the error:
> 
> AttributeError: 'UbuntuDistroInfo' object has no attribute 'get_all'
> 
> This is very confusing, since the code from this (from the distro-info
> library) is essentially:
> 
> class DistroInfo:
> def get_all(self, result="codename"):
> ...
> 
> class UbuntuDistroInfo(DistroInfo):
> def __init__(self):
> super().__init__("Ubuntu")
> 
> i.e. I can't see any reason why the `di` object would not have a get_all()
> method. Running software-properties and various test scripts on a bionic VM
> I haven't been able to reproduce any such issues.
> 
> Theories that have been proposed:
> - The user is actually running an older version of disto-info that was from
> before this method was added (in distro-info 0.15).
>   - Seems unlikely since there are error cases in which the install was
> done with a bionic image.
> - The user has a locally installed version of distro-info.
>   - Stacktraces shows the system installed version, the appropriate version
> of the binary installed, no PYTHONPATH set.
> - A .pyc is being used that doesn't match the source.
>   - Since installs have been from bionic media, it seems impossible that an
> older .pyc could be created.
> - Other code has removed this method.
>   - Method is being accessed immediately, no other code seen in source that
> could do this.
> - Memory corruption.
>  - Too many cases for error to be random, not seeing other similar issues.
> 
> Does anyone have any ideas about what might be going on?

Could it be an old pip-installed version of distro-info?
(C.f. LP: #1848829, #1874250)

Bryce

> --Robert
> 
> [1] https://launchpad.net/ubuntu/+source/software-properties/0.96.24.32.18
> [2]
> https://errors.ubuntu.com/problem/477791e8cc662f8dca46050bb638e273028a522f

> -- 
> ubuntu-devel mailing list
> ubuntu-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: 
> https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel


-- 
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel