[arch-general] update to calibre-0.9.15-1
during today's update from calibre 0.9.14-1 -> 0.9.15-1, pacman complained that all packages under /usr/lib/calibre/calibre were already existing in the filesystem, and therefore the update couldn't be performed. for some reason "pacman -Qo" claimed that all these files didn't belong to any package. only after moving them from that location somewhere else could the new version of calibre be installed. looks to me as if this is some packaging error. don't know enough about arch packaging though to say what exactly went wrong... -- phani.
Hi, Am 18.01.2013 13:59, schrieb phani:
during today's update from calibre 0.9.14-1 -> 0.9.15-1, pacman complained that all packages under /usr/lib/calibre/calibre were already existing in the filesystem, and therefore the update couldn't be performed.
Not sure what was going over on your side, however the update went smoothly for me.
looks to me as if this is some packaging error. don't know enough about arch packaging though to say what exactly went wrong...
Looking at the changes done to package, I don't think that this is a packaging error, see [1]. Had you issues with other packages, too? [1]: https://projects.archlinux.org/svntogit/community.git/log/trunk?h=packages/c... Best regards, Karol Babioch
On Fri, 18 Jan 2013 18:52:08 +0530, Karol Babioch <karol@babioch.de> wrote:
Looking at the changes done to package, I don't think that this is a packaging error, see [1].
Had you issues with other packages, too?
no, everything else went smooth; and after moving those calibre files, so went the calibre update. didn't make any manual changes to files under /usr/lib/~ in a long time, nor did my computer crash recently. don't remember if my internet connection went dead during the first attempt to update, that happens frequently. but i don't see how that would mess up things -- unless everything is downloaded & checked, pacman doesn't do anything. weird, but nothing to loose sleep over. some freak accident... -- phani.
don't remember if my internet connection went dead during the first attempt to update, that happens frequently. but i don't see how that would mess up things -- unless everything is downloaded & checked, pacman doesn't do anything.
It is common issue when you run as root python program without precompiled pyc/pyo, now pyc/pyo provided by package and conflict with autogenerated.
On Fri, 18 Jan 2013 19:14:54 +0530, Alexander Bashmakov <alex.teorver@gmail.com> wrote:
It is common issue when you run as root python program without precompiled pyc/pyo, now pyc/pyo provided by package and conflict with autogenerated.
the only thing i ran as root was pacman; certainly not calibre. -- phani.
On Fri, Jan 18, 2013 at 2:22 PM, Karol Babioch <karol@babioch.de> wrote:
Hi,
Am 18.01.2013 13:59, schrieb phani:
during today's update from calibre 0.9.14-1 -> 0.9.15-1, pacman complained that all packages under /usr/lib/calibre/calibre were already existing in the filesystem, and therefore the update couldn't be performed.
Not sure what was going over on your side, however the update went smoothly for me.
looks to me as if this is some packaging error. don't know enough about arch packaging though to say what exactly went wrong...
Looking at the changes done to package, I don't think that this is a packaging error, see [1].
No one that hasn't touched the calibre software will suffer from the existing files. Python (which creates those files when stuff is used) packages now provide compiled bytecode which also saves the users' power, and I think it's a great feature. The downside is that all the *.pyc/*.pyo stuff is in many cases already on disk and it was actually the third hunk in [1] that brought in the change. Look up the bug report [2] that hunk is referring to, which is relevant in this case. There were other such updates in the past for me - and I believe I used some for/rm loop in bash instead of pacman --force. I think top note in [3] was added at some point, which will both impose further cleanness in python packaging and confusion on not-very-savant users in the future. I also agree with pacman in that it should not handle such cases, because that's what using arch means to me. Alternatively, you could have informed yourself about what *.pyc/*.pyo files are and answered your own question. cheers! mar77i [1] https://projects.archlinux.org/svntogit/community.git/log/trunk?h=packages/c... [2] https://bugs.archlinux.org/task/33392 [3] https://wiki.archlinux.org/index.php/Python_Package_Guidelines#Notes
On Fri, 18 Jan 2013 19:48:37 +0530, Martti Kühne <mysatyre@gmail.com> wrote:
No one that hasn't touched the calibre software will suffer from the existing files.
i haven't touched the calibre software -- if that means done any manual changes to it's files. if you mean "used the calibre software," that i did.
Python (which creates those files when stuff is used) packages now provide compiled bytecode which also saves the users' power, and I think it's a great feature. The downside is that all the *.pyc/*.pyo stuff is in many cases already on disk and it was actually the third hunk in [1] that brought in the change. Look up the bug report [2] that hunk is referring to, which is relevant in this case. There were other such updates in the past for me - and I believe I used some for/rm loop in bash instead of pacman --force.
i just moved the not-owned files somewhere else. wouldn't think of using --force unless i knew exactly what was going on.
I think top note in [3] was added at some point, which will both impose further cleanness in python packaging and confusion on not-very-savant users in the future. I also agree with pacman in that it should not handle such cases, because that's what using arch means to me.
Alternatively, you could have informed yourself about what *.pyc/*.pyo files are and answered your own question.
"RTFM" makes sense to me. "study python packaging before asking the list" does not. -- phani.
On Sat, 19 Jan 2013 03:16:58 +0530, phani <listmail@phanisvara.com> wrote:
Alternatively, you could have informed yourself about what *.pyc/*.pyo files are and answered your own question.
"RTFM" makes sense to me. "study python packaging before asking the list" does not.
that probably came across a bit harsh or ungrateful. i do thank you for pointing me into the right direction to understand what's going on. (i've never used python in my life, so these things are new to me.) -- phani.
participants (4)
-
Alexander Bashmakov
-
Karol Babioch
-
Martti Kühne
-
phani