The Distribution line is unnecessary since it only has meaning on 10.4 and 10.5 which are unsupported.
numexpr-py currently only has 2.7 and 3.3 variants so pytables has to be restricted to the same until numexpr-py gets updated.
A dependency on hdf5.8-shlibs | hdf5.7-shlibs is against fink policy since it makes package building non-deterministic. You should just use hdf5.8-shlibs since it's newest and the only one available on 10.9. Same is true of hdf5.8 | hdf5.7 of course.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Shouldn't push submissions this late!
I cleaned this up, and fixed also the suggested minimum versions for some other dependencies.
The (trivial) update to numexpr-py is submitted in [#4397]; I also suggested to keep its py32 version for now, as that seemed to be the consensus on fink-devel.
Attached .info is restricted to 2.7 and 3.3 for the time being.
Preparing to replace pytables-py27 2.4.0-1 (using .../pytables-py27_3.1.1-1_darwin-x86_64.deb) ...
Unpacking replacement pytables-py27 ...
Setting up pytables-py27 (3.1.1-1) ...
update-alternatives: error: alternative ptdump can't be slave of pt2to3: it is a slave of nctoh5
dpkg: error processing pytables-py27 (--install):
subprocess installed post-installation script returned error exit status 2
Errors were encountered while processing:
pytables-py27
### execution of /sw/bin/dpkg-lockwait failed, exit code 1
Apparently, update-alternatives doesn't like the 'master' being changed during an upgrade. Do note: I am using TheSin's new dpkg branch which has a less forgiving update-alternatives.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Think this must be dpkg-specific then, direct update from 2.3.1 worked for me (could only test 10.8+10.9 though).
No idea if this can be addressed in the PostInstScript; but a full remove ipython-py27 before the update should work around this, right?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Yes, manually removing the old pytables (not ipython) first let the new pytables-3.1.1 install. But I had to do a force remove because I had pandas-py installed (which deps on pytables). If it is a new-dpkg issue, this is not necessarily a 'must fix now' issue, but is liable to come up again in the future with other packages. A general method to deal with this situation on the .info side should be found.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
You're welcome to take it over!
There are a couple of issues though.
The Distribution line is unnecessary since it only has meaning on 10.4 and 10.5 which are unsupported.
numexpr-py currently only has 2.7 and 3.3 variants so pytables has to be restricted to the same until numexpr-py gets updated.
A dependency on hdf5.8-shlibs | hdf5.7-shlibs is against fink policy since it makes package building non-deterministic. You should just use hdf5.8-shlibs since it's newest and the only one available on 10.9. Same is true of hdf5.8 | hdf5.7 of course.
Shouldn't push submissions this late!
I cleaned this up, and fixed also the suggested minimum versions for some other dependencies.
The (trivial) update to numexpr-py is submitted in [#4397]; I also suggested to keep its py32 version for now, as that seemed to be the consensus on fink-devel.
Attached .info is restricted to 2.7 and 3.3 for the time being.
Related
Package Submissions:
#4397This broke for me on 10.7:
Apparently, update-alternatives doesn't like the 'master' being changed during an upgrade. Do note: I am using TheSin's new dpkg branch which has a less forgiving update-alternatives.
Think this must be dpkg-specific then, direct update from 2.3.1 worked for me (could only test 10.8+10.9 though).
No idea if this can be addressed in the PostInstScript; but a full remove ipython-py27 before the update should work around this, right?
Yes, manually removing the old pytables (not ipython) first let the new pytables-3.1.1 install. But I had to do a force remove because I had pandas-py installed (which deps on pytables). If it is a new-dpkg issue, this is not necessarily a 'must fix now' issue, but is liable to come up again in the future with other packages. A general method to deal with this situation on the .info side should be found.