Bug#868379: calibre breaks with recent python-dateutil
Hi,
On Sun, Jul 16, 2017 at 05:12:28PM +0200, Manolo Díaz wrote:
> On Sunday, 16 Jul 2017 at 11:27 UTC
> Guido Günther wrote:
>
> > Hi,
> > On Sun, Jul 16, 2017 at 11:22:47AM +0200, Manolo Díaz wrote:
> > > On Sunday, 16 Jul 2017 at 09:05 UTC
> > > Guido Günther wrote:
> > >
> > > > Hi,
> > > > control: affects -1 calibre
> > >
> > > [...]
> > >
> > > > Starts here without problems. Does
> > > >
> > > > python -c "from six.moves import _thread"
> > > >
> > > > work for you? If it does not work your python-six is broken. Maybe you
> > > > have a local version of six lying around?
> > > > Cheers,
> > > > -- Guido
> > >
> > > Hi,
> > >
> > > python -c "from six.moves import _thread" does apparently nothing and
> > > exits with 0. Is this expected?
> > >
> >
> > Yept. That's how it should be.
> >
> > > The version of python-six on my system is 1.10.0-4 (current testing)
> > > and if it was broken I think calibre wouldn't start with only
> > > downgrading python-dateutil.
> >
> > It might because the old python-dateutil version might not have used it.
> > Can you do a
> >
> >strace -f -s2048 /usr/bin/calibre
> >
> > and attach this to the bugreport please.
> > Cheers,
> > -- Guido
>
> Of course. It's attached.
You have six.pyc in the calibre directory that's tripping up things:
open("/usr/lib/calibre/six.pyc", O_RDONLY) = 9
can you remove that and try again?
Cheers,
-- Guido
Bug#868379: calibre breaks with recent python-dateutil
On Sunday, 16 Jul 2017 at 15:24 UTC
Guido Günther wrote:
> Hi,
> On Sun, Jul 16, 2017 at 05:12:28PM +0200, Manolo Díaz wrote:
> > On Sunday, 16 Jul 2017 at 11:27 UTC
> > Guido Günther wrote:
> >
> > > Hi,
> > > On Sun, Jul 16, 2017 at 11:22:47AM +0200, Manolo Díaz wrote:
> > > > On Sunday, 16 Jul 2017 at 09:05 UTC
> > > > Guido Günther wrote:
> > > >
> > > > > Hi,
> > > > > control: affects -1 calibre
> > > >
> > > > [...]
> > > >
> > > > > Starts here without problems. Does
> > > > >
> > > > > python -c "from six.moves import _thread"
> > > > >
> > > > > work for you? If it does not work your python-six is broken. Maybe you
> > > > > have a local version of six lying around?
> > > > > Cheers,
> > > > > -- Guido
> > > >
> > > > Hi,
> > > >
> > > > python -c "from six.moves import _thread" does apparently nothing and
> > > > exits with 0. Is this expected?
> > > >
> > >
> > > Yept. That's how it should be.
> > >
> > > > The version of python-six on my system is 1.10.0-4 (current testing)
> > > > and if it was broken I think calibre wouldn't start with only
> > > > downgrading python-dateutil.
> > >
> > > It might because the old python-dateutil version might not have used it.
> > > Can you do a
> > >
> > >strace -f -s2048 /usr/bin/calibre
> > >
> > > and attach this to the bugreport please.
> > > Cheers,
> > > -- Guido
> >
> > Of course. It's attached.
>
> You have six.pyc in the calibre directory that's tripping up things:
>
> open("/usr/lib/calibre/six.pyc", O_RDONLY) = 9
>
> can you remove that and try again?
> Cheers,
> -- Guido
It works. After removing /usr/lib/calibre/six.pyc calibre works again.
So you are right, it's not a python-datetime bug.
Thank you.
Regards,
--
Manolo Díaz
Bug#868379: calibre breaks with recent python-dateutil
control: reassign -1 calibre
Hi,
On Sun, Jul 16, 2017 at 05:33:53PM +0200, Manolo Díaz wrote:
> On Sunday, 16 Jul 2017 at 15:24 UTC
> Guido Günther wrote:
>
> > Hi,
> > On Sun, Jul 16, 2017 at 05:12:28PM +0200, Manolo Díaz wrote:
> > > On Sunday, 16 Jul 2017 at 11:27 UTC
> > > Guido Günther wrote:
> > >
> > > > Hi,
> > > > On Sun, Jul 16, 2017 at 11:22:47AM +0200, Manolo Díaz wrote:
> > > > > On Sunday, 16 Jul 2017 at 09:05 UTC
> > > > > Guido Günther wrote:
> > > > >
> > > > > > Hi,
> > > > > > control: affects -1 calibre
> > > > >
> > > > > [...]
> > > > >
> > > > > > Starts here without problems. Does
> > > > > >
> > > > > > python -c "from six.moves import _thread"
> > > > > >
> > > > > > work for you? If it does not work your python-six is broken. Maybe
> > > > > > you
> > > > > > have a local version of six lying around?
> > > > > > Cheers,
> > > > > > -- Guido
> > > > >
> > > > > Hi,
> > > > >
> > > > > python -c "from six.moves import _thread" does apparently nothing and
> > > > > exits with 0. Is this expected?
> > > > >
> > > >
> > > > Yept. That's how it should be.
> > > >
> > > > > The version of python-six on my system is 1.10.0-4 (current testing)
> > > > > and if it was broken I think calibre wouldn't start with only
> > > > > downgrading python-dateutil.
> > > >
> > > > It might because the old python-dateutil version might not have used it.
> > > > Can you do a
> > > >
> > > >strace -f -s2048 /usr/bin/calibre
> > > >
> > > > and attach this to the bugreport please.
> > > > Cheers,
> > > > -- Guido
> > >
> > > Of course. It's attached.
> >
> > You have six.pyc in the calibre directory that's tripping up things:
> >
> > open("/usr/lib/calibre/six.pyc", O_RDONLY) = 9
> >
> > can you remove that and try again?
> > Cheers,
> > -- Guido
>
> It works. After removing /usr/lib/calibre/six.pyc calibre works again.
> So you are right, it's not a python-datetime bug.
So calibre should clean up *.pyc files (probably from ancient
installations) in /usr/lib/calibre.
Cheers,
-- Guido
Bug#868379: calibre breaks with recent python-dateutil
On Sunday, 16 Jul 2017 at 15:59 UTC
Guido Günther wrote:
> control: reassign -1 calibre
>
> Hi,
> On Sun, Jul 16, 2017 at 05:33:53PM +0200, Manolo Díaz wrote:
> > On Sunday, 16 Jul 2017 at 15:24 UTC
> > Guido Günther wrote:
> >
> > > Hi,
> > > On Sun, Jul 16, 2017 at 05:12:28PM +0200, Manolo Díaz wrote:
> > > > On Sunday, 16 Jul 2017 at 11:27 UTC
> > > > Guido Günther wrote:
> > > >
> > > > > Hi,
> > > > > On Sun, Jul 16, 2017 at 11:22:47AM +0200, Manolo Díaz wrote:
> > > > > > On Sunday, 16 Jul 2017 at 09:05 UTC
> > > > > > Guido Günther wrote:
> > > > > >
> > > > > > > Hi,
> > > > > > > control: affects -1 calibre
> > > > > >
> > > > > > [...]
> > > > > >
> > > > > > > Starts here without problems. Does
> > > > > > >
> > > > > > > python -c "from six.moves import _thread"
> > > > > > >
> > > > > > > work for you? If it does not work your python-six is broken.
> > > > > > > Maybe you
> > > > > > > have a local version of six lying around?
> > > > > > > Cheers,
> > > > > > > -- Guido
> > > > > >
> > > > > > Hi,
> > > > > >
> > > > > > python -c "from six.moves import _thread" does apparently nothing
> > > > > > and
> > > > > > exits with 0. Is this expected?
> > > > > >
> > > > >
> > > > > Yept. That's how it should be.
> > > > >
> > > > > > The version of python-six on my system is 1.10.0-4 (current testing)
> > > > > > and if it was broken I think calibre wouldn't start with only
> > > > > > downgrading python-dateutil.
> > > > >
> > > > > It might because the old python-dateutil version might not have used
> > > > > it.
> > > > > Can you do a
> > > > >
> > > > >strace -f -s2048 /usr/bin/calibre
> > > > >
> > > > > and attach this to the bugreport please.
> > > > > Cheers,
> > > > > -- Guido
> > > >
> > > > Of course. It's attached.
> > >
> > > You have six.pyc in the calibre directory that's tripping up things:
> > >
> > > open("/usr/lib/calibre/six.pyc", O_RDONLY) = 9
> > >
> > > can you remove that and try again?
> > > Cheers,
> > > -- Guido
> >
> > It works. After removing /usr/lib/calibre/six.pyc calibre works again.
> > So you are right, it's not a python-datetime bug.
>
> So calibre should clean up *.pyc files (probably from ancient
> installations) in /usr/lib/calibre.
> Cheers,
> -- Guido
It would help, but I think it isn't enough. Given the following
sequence calibre would have failed to start.
- upgrade or reinstall calibre (all *.pyc are cleaned up)
- run calibre as root (they are created again)
- upgrade python-dateutil.
Thanks again.
Regards,
--
Manolo Díaz
Bug#868379: calibre breaks with recent python-dateutil
Hello Manolo, Manolo Díaz [2017-07-16 18:36 +0200]: > > So calibre should clean up *.pyc files (probably from ancient > > installations) in /usr/lib/calibre. > > Cheers, > > -- Guido > > It would help, but I think it isn't enough. Given the following > sequence calibre would have failed to start. > > - upgrade or reinstall calibre (all *.pyc are cleaned up) > - run calibre as root (they are created again) > - upgrade python-dateutil. I don't think that there's much that a Python application package can do about it. Cleaning up the pyc files on package upgrades should alleviate this (I'll add that to the packaging), but I figure the mch better solution here is "don't run calibre as root". Martin
Processed: Re: Bug#868379: calibre breaks with recent python-dateutil
Processing control commands: > reassign -1 calibre Bug #868379 [python-dateutil] python-dateutil: makes calibre fail to start Bug reassigned from package 'python-dateutil' to 'calibre'. No longer marked as found in versions python-dateutil/2.6.0-1. Ignoring request to alter fixed versions of bug #868379 to the same values previously set -- 868379: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=868379 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems
