It seems I am running into troubles with my fink installation on a
case-sensitive volume again,
but I don't quite understand the issue. perl5162-core builds ok so far, but the
package fails to
install with
Validating package file
/sw/fink/dists/stable/main/finkinfo/languages/perl5162.info...
Hi Dan,
thanks for the pointer, this seems to be a bigger can of worms than I've
thought!
It would seem, since the .deb contains both the lower- and uppercase version
dirs with their content,
it should install just fine on a case-sensitive filesystem, but apparently
dpkg-deb is still
On 19.11.2013, at 7:47PM, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
The 'missing' perl5.16 should be fixed in fink-0.36.1, released two days ago.
Fink was incorrectly assuming that all perls were perlX.Y, but this is only
the case for system-perl, while fink's perls have
On 19.11.2013, at 8:36PM, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
On 11/19/2013 1:55 PM, Derek Homeier wrote:
On 19.11.2013, at 7:47PM, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
The 'missing' perl5.16 should be fixed in fink-0.36.1, released two days
ago
On 20.11.2013, at 2:29AM, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
On 11/19/2013 8:04 PM, Derek Homeier wrote:
But a lot of modules fail to build with this missing dependency on
cc_runtime.h.
There is no cc_runtime.h of either capitalisation in
/sw/lib/perl5-core/5.16.2
throwing in a Python3 flavour of h5py, all built and tested (fink -m)
with
python2.6-3.3 under 10.5 and 10.8 (except for a couple of errors with
python3.1).
Cheers,
Derek
--
Derek Homeier
On 20.03.2014, at 4:20PM, Daniel Johnson daniel.johnso...@gmail.com wrote:
I agree. There's no reason to keep 2.6 around as 2.7 is fully backward
compatible.
Just an aside, anyone aware of any major Linux distros that still only provide
2.6?
This was an issue for many packages like numpy to
On 23.03.2014, at 9:50PM, Daniel Johnson daniel.johnso...@gmail.com wrote:
Does anyone know btw if 3.4 already comes with full setuptools support
included?
It has a %i/bin/easy_install-3.4, and I was able to build and test
coverage, nose
and numpy without installing it, so maybe a change
Hi,
there seems to be some mixup of the source files and checksums for pyqt4:
The checksum of the file PyQt-mac-gpl-4.10.4.tar.gz of package
pyqt4-py27-4.10.4-1 is incorrect. The most likely cause
for this is a corrupted or incomplete download
Expected: 20940f6b2c957269cdd02d0efe537515
Actual:
Hi,
found there is a missing
TestDepends: six-py%type_pkg[python]
in dateutil-py-2.2-1.
Cheers,
Derek
--
___
Fink-devel mailing
fink update sip-py33-bin
Information about 9023 packages read in 1 seconds.
The following package will be installed or updated:
sip-py33-bin
Reading buildlock packages...
All buildlocks accounted for.
On 05.04.2014, at 11:44PM, Daniel Johnson daniel.johnso...@gmail.com wrote:
Well, not quite. The patch doesn't apply. Sigh. I'll have to work on it some
more. I don't use the -x11 package so it doesn't get a lot of attention and I
don't want to spend hours building qt4-x11.
It builds with
On 06.04.2014, at 5:58PM, Daniel Macks dma...@netspace.org wrote:
found there is a missing
TestDepends: six-py%type_pkg[python]
in dateutil-py-2.2-1.
Fixed (sip-pyXXX not six-pyXXX).
hmm, I had gotten this error:
sudo -u fink-bld [ENV] sh -c /tmp/fink.67Gmy
/tmp/fink.3vLAc
On 06.04.2014, at 9:33PM, Derek Homeier de...@astro.physik.uni-goettingen.de
wrote:
On 06.04.2014, at 5:58PM, Daniel Macks dma...@netspace.org wrote:
found there is a missing
TestDepends: six-py%type_pkg[python]
in dateutil-py-2.2-1.
Fixed (sip-pyXXX not six-pyXXX).
hmm, I
Hi,
h5py-py no longer builds with hdf5.8_1.8.13 due to an obsolete macro (meaning
that the
current version no longer works either unless enforcing
HDF5_DISABLE_VERSION_CHECK).
I've submitted a version patched with the appropriate changeset from github
master (#4416).
Cheers,
On 21.06.2014, at 5:44PM, Kurt Schwehr kurtschw...@yahoo.com wrote:
:(. I saw the same issue recently, but have not yet had time to investigate
-Kurt (maintainer of the h5py fink package)
Apparently the H5FD_MPIPOSIX macro was not used anyway, so all it took
was commenting it out; so
On 22.06.2014, at 1:54AM, Kurt Schwehr goat...@users.sf.net wrote:
Thanks for finding the issue. Patch files drive me crazy, so I added the two
changes as a patch script with two perl one liners.
Thanks for the quick fix. Strangely, it did not work on the quite fresh fink
installation
of my
On 23 Jun 2014, at 01:43 pm, Jesse Alama jesse.al...@gmail.com wrote:
Emacs depends on the pkg 'passwd-games' so that user should normally be set
up on the system. Try 'fink reinstall passwd-games' and select yes when
asked to install the user.
Thanks, that does seem to have made a
Hi,
forwarding this here in case there are other Fink packages where this has
become an issue
(the package in question is on
https://sourceforge.net/p/fink/package-submissions/4396/):
Begin forwarded message:
I had a look at the new release and actually ran into a problem right away
Hi,
FYI building aquaterm with package validation fails for me on a *case
sensitive* file system:
Validating .deb dir /scratch.noindex/fink.build/root-aquaterm-dev-1.1.1-3...
Error: Pathname collision on case-insensitive filesystems
Offending file: /sw/share/aquaterm/adapters/c/ReadMe
Hi Kevin,
Fink Developers - I’m not sure of the best way to fix the problem. I
searched man tar, but don’t see any options to deal with this. I could
create a new tarball with file names that don’t collide, load that on the
web somewhere, and change the source in the info to point to
Hi Kevin,
I committed an aquaterm update which I think corrects the problem. But, the
original package validated for me, so I cannot be sure my fix will pass the
validator.
yes, no way really to find if it breaks on a case-insensitive fs without
building on a case-sensitive one…
But
Hi,
the validation step fails when built with update -m:
Validating .deb dir
/scratch.noindex/fink.build/root-libwpg-0.2-shlibs-0.2.2-2...
Error:
/scratch.noindex/fink.build/root-libwpg-0.2-shlibs-0.2.2-2/sw/lib/libwpg-0.2.2.dylib
appears to have been linked using a flat namespace.
If
--
Derek Homeier Centre de Recherche Astrophysique de Lyon
ENS Lyon 46, Allée d'Italie
69364 Lyon Cedex 07, France +33 47272-8894
Hi Kurt,
I found some issues with the build_tests run added to make the test suite for
nose 1.3.4 work:
setup.py build_tests makes the test run with python3.x, but breaks it for
python2.7 (setup.py test actually works there).
The same workaround seems to install an additional nose.egg-info in
On 6 Feb 2015, at 4:44 pm, Derek Homeier de...@astro.physik.uni-goettingen.de
wrote:
I found some issues with the build_tests run added to make the test suite for
nose 1.3.4 work:
setup.py build_tests makes the test run with python3.x, but breaks it for
python2.7 (setup.py test actually
Hi,
I think the Usage Notes for the HDF5.9 Fortran libraries are outdated or
mistakenly
taken over from the -oldapi variant (pointing to the %p/opt/hdf5v1.6 subdirs).
Also the InstallScript is removing all executables including the Fortran
wrapper h5fc
which is not provided by hdf5.9.
I suggest
On 18 Feb 2015, at 12:44 am, Daniel Macks dma...@netspace.org wrote:
as the entire test suite fails with an import error with 1.8 (apparently
crashes the
interpreter, thus the failure was not caught be the original Test field).
I think the usual pattern for TestScript is:
some command
Hi,
following up on my recent question how to best get package updates into the
distribution as a non-core developer, is the full package tree going to be
migrated
to github at some point? Then it would of course just take a pull request.
In the meantime, I am attaching the update for Astropy
Hi Jack,
The bottleneck-py package needs updated to the 1.0.0 source in
order to pass 'fink -m' against the newer numpy 1.9.x release…
thanks for the patch; the Depends then also need to be updated to
Depends: python%type_pkg[python], numpy-py%type_pkg[python] (= 1.9.0-1)
as the entire
Hi,
the nightfall update is still failing with name conflicts on case-sensitive HFS
(and is silently losing 2 of the files in question on insensitive ones).
This can be fixed with a
TarFilesRename: bh_virB.dat:bh_virBoldZ.dat bh_virV.dat:bh_virVoldZ.dat
see also the last update to
On 18 Mar 2015, at 07:26 pm, Daniel Macks dma...@netspace.org wrote:
On Wed, 18 Mar 2015 09:38:32 -0700, Alexander Hansen
alexanderk.han...@gmail.com wrote:
On Mar 18, 2015, at 9:08 AM, Derek Homeier
de...@astro.physik.uni-goettingen.de wrote:
Hi,
the nightfall update is still
Hi,
building vtk510-5.10.1-1 with -m fails the package check due to two entries
missing in the Shlibs list:
147a148
%p/lib/%N/libLSDyna.5.10.dylib5.10.0 %n (=
5.10.0-1)
231a233
%p/lib/%N/libvtkhdf5_hl.1.8.5.dylib 1.8.5 %n (=
5.10.0-1)
Hanspeter, Alexander,
On 25 Feb 2015, at 5:05 am, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
following up on my recent question how to best get package updates into the
distribution as a non-core developer, is the full package tree going to be
migrated
to github at some
On 9 Apr 2015, at 7:27 pm, Alexander Hansen alexanderk.han...@gmail.com wrote:
It appears to need Apple’s Legacy Java-1.6, and thus system-java16*. You can
get that:
https://support.apple.com/downloads/DL1572/en_US/JavaForOSX2014-001.dmg
The build worked for me when I moved my Java-1.8
On 6 Apr 2015, at 1:03 am, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
building vtk510-5.10.1-1 with -m fails the package check due to two entries
missing in the Shlibs list:
147a148
%p/lib/%N/libLSDyna.5.10.dylib 5.10.0 %n (=
5.10.0-1)
231a233
On 6 Apr 2015, at 1:03 am, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
building vtk510-5.10.1-1 with -m fails the package check due to two entries
missing in the Shlibs list:
147a148
%p/lib/%N/libLSDyna.5.10.dylib 5.10.0 %n (=
5.10.0-1)
On 6 Apr 2015, at 1:03 am, Hanspeter Niederstrasser f...@snaggledworks.com
wrote:
building vtk510-5.10.1-1 with -m fails the package check due to two entries
missing in the Shlibs list:
147a148
%p/lib/%N/libLSDyna.5.10.dylib 5.10.0 %n (=
5.10.0-1)
231a233
Hi Jack,
the 1.8 versions of Open-MPI have a couple of new files with conflicting
filenames,
causing the build on case-sensitive HFS+ to fail (and probably some of those
files
to be lost on case-insensitive ones. The vtCC file versions should be added to
the
case-sensitive fix, I think:
Hi,
this package does not update due to a non-existing dependency in the -dev
Splitoff:
< Depends: graphicsmagick1321-q32-shlibs (= %V)
---
> Depends: graphicsmagick1322-q32-shlibs (= %V)
Cheers,
Derek
Hi,
building hdf5-bin_1.8.16 fails with
env LANG=C LC_ALL=C dpkg-deb -b root-hdf5-bin-1.8.16-1
/sw/fink/10.10/stable/main/binary-darwin-x86_64/sci
/sw/bin/dpkg-deb: parse error, in file `root-hdf5-bin-1.8.16-1/DEBIAN/control'
near line 10 package `hdf5-bin':
`Replaces' field, syntax error
Hi,
on a case-sensitive HFS+ build I am getting the following error
Validating .deb dir
/scratch.noindex/fink.build/root-imagemagick6.q16-doc-6.9.3.2-2...
Error: Pathname collision on case-insensitive filesystems
Offending file: /sw/share/doc/ImageMagick-6/www/api/image.html
which
On 11 Feb 2016, at 1:25 am, Alexander Hansen
wrote:
>
> urllib.error.URLError: certificate verify failed (_ssl.c:645)>
>
> During handling of the above exception, another exception occurred:
>
> Traceback (most recent call last):
> File
11 Feb 2016, at 1:22 am, Alexander Hansen <alexanderk.han...@gmail.com>
> wrote:
>
>> On Feb 10, 2016, at 15:55, Derek Homeier
>> <de...@astro.physik.uni-goettingen.de> wrote:
>>
>>
>>> On 11 Feb 2016, at 12:35 am, Alexander Hansen <alexa
On 12 Feb 2016, at 8:42 pm, Derek Homeier
<de...@astro.physik.uni-goettingen.de> wrote:
>
> On 11 Feb 2016, at 2:12 pm, Kurt Schwehr <kurtschw...@yahoo.com> wrote:
>>
>> Sorry, I'm totally not following how this should be fixed. I'm feeling
>> rather brai
On 16 Feb 2016, at 2:35 pm, Kurt Schwehr wrote:
>
> If you have a working setup for matplotlib, can you commit it?
>
I’ve tested and built these on 10.10 and 10.11; can submit them to the package
tracker.
Cheers,
Derek
> On 11 Feb 2016, at 12:35 am, Alexander Hansen
> wrote:
>
>> when building the updated matplotlib package in maintainer mode, after the
>> tests are run I am
>> still getting the dreaded verification error for the .deb:
>>
>> Error: Compiled python module points
Hi,
when building the updated matplotlib package in maintainer mode, after the
tests are run I am
still getting the dreaded verification error for the .deb:
Error: Compiled python module points to fink build dir.
Uncommenting the cleanup line does not help, because after a few Fails and
Hi Kurt,
> On 19 Jan 2017, at 9:18 pm, Kurt Schwehr wrote:
>
> yaml-py is updated. If you want to take over matplotlib (or any of my other
> packages), be my guest
>
thanks; I have now uploaded the matplotlib-2.0 update. If you wish, I can take
it over as maintainer,
HI,
when installing the py36 version of the new numpy version (thanks for the swift
update!),
it refuses to build with tests enabled, as there is no py36 version of nose
available yet.
nose-py-1.3.7 builds out of the box on 10.11 and 10.12 by just adding 3.6 to
the Type
field, and numpy-py36
have also just submitted two package requirements for matplotlib-2.0.0 to the
tracker,
which you can take over as maintainer or leave me in, whatever you prefer.
Cheers,
Derek
>
> On Tuesday, January 17, 2017 7:03 AM, Derek Homeier
> <de...@ast
Hi Daniel,
subject already tells most of the story: the capitalisation workaround in
openssl110-dev incorrectly copies
the MD5.3 manpage over in lowercase, which fails on a case-sensitive FS. I
think this should fix it:
diff -Nurd
Hi,
building the package with ‘-m’ fails on a missing Shlibs entry (the other
boost1.63 variants build fine):
Validating .deb dir
/scratch.noindex/fink.build/root-boost1.63-nopython-1.63.0-1...
Error: package contains the shared library
/sw/opt/boost-1_63/lib/libboost_fiber-mt.dylib
On 14 Jan 2017, at 3:14 pm, Hanspeter Niederstrasser
wrote:
>
> This has been fixed by disabling the boost_filter library. If someone
> requests it in the future (or we stop supporting 10.9 or 10.10), I'll
> re-enable it.
Tested and found working here on 10.11 and
> On 11 Jan 2017, at 11:46 am, Hanspeter Niederstrasser
> wrote:
>
>> building the package with ‘-m’ fails on a missing Shlibs entry (the other
>> boost1.63 variants build fine):
>>
>> Validating .deb dir
>>
On 11 Jan 2017, at 12:28 pm, Hanspeter Niederstrasser
wrote:
>
>> + LIBDIR2=
>> + ./b2 -d2 'cxxflags=-std=c++11 -MD -I/sw/include' -j4 --without-python
>> --prefix=/sw --libdir=/sw/opt/boost-1_63/lib
>> --includedir=/sw/opt/boost-1_63/include/boost
[cc’ing Daniel as OpenSSL maintainer]
> On 4 Jan 2017, at 12:47 pm, Hanspeter Niederstrasser <f...@snaggledworks.com>
> wrote:
>
> On 1/2/17 9:19 AM, Derek Homeier wrote:
>> Hi,
>>
>> I want to check if I am doing something very stupid here, since I am un
Hi,
I want to check if I am doing something very stupid here, since I am unable to
properly
use any apps linking to openssl100-shlibs (among others wget and python) since
approximately
the update to openssl-1.0.2, as it refuses to accept almost any host
certificate:
ariel:15579> curl -O
58 matches
Mail list logo