Hi,
It's just so you don't have to maintain your own version of the
'sconstruct' script, which is the equivalent of the 'makefile'. But
if you have to define a specific Python version and the
FINK_INSTALL_DIR variable, then it would be good to have this in the
script. For FINK_INSTALL_DIR and the commented out *.pyc
compilation, these are only used by the 'install' and 'uninstall'
scons targets which are not necessary under fink. This is a side
question, but does the fink package contain the *.pyc files, or are
these created afterwards? For the Python version, I need to do this
anyway for other operating systems for the shift to Python 3.x (in the
very distant future) and it would be useful for my own testing of
relax against the numerous Python versions.
For the test-suite, as soon as you post the bug reports, I'll fix
these. From memory they are only affecting the optimisation to the
precision of the 7th or 8th decimal point, so I just have to catch
this (I basically only need the numbers printed at the bottom of the
test). But for relax usage this is a non-issue.
Cheers,
Edward
2009/8/18 Jack Howarth <howarth@xxxxxxxxxxxxxxxx>:
Edward,
I'll have to check with the other fink developers as I've never
heard of integration of the fink packaging directly into the makefiles.
It really isn't a big issue since the user needs fink installed anyway.
All they need to do is execute one of the following commands...
fink install relax-py24
fink install relax-py25
fink install relax-py26
A more productive issue may be to look at the few failures I am
seeing with 'relax --test-suite' when compiled and executed on
x86_64 fink. I suspect most are minor rounding differences in
the output. When I get a chance later this week, I'll start to
file those in your bugzilla.
Jack
On Tue, Aug 18, 2009 at 03:38:13PM +0200, Edward d'Auvergne wrote:
Thank you very much! I've applied this patch to the 1.3 development
branch (this is 'trunk', but note there is also a 1.2 'trunk') as
revision r9332. For the other changes, I would like to incorporate
these as permanent changes as well. Could you detail all the steps
you perform to generate the Fink package? Maybe we could modify the
'sconstruct' script to do most of this automatically when typing
"scons binary_dist" on Mac OS X. No one else uses this scons target
on Mac OS X, therefore we could make it Fink specific. Could you
also explain these other changes to the relax code, line by line? For
selecting a specific python version, I'm thinking that we can do
something like:
$ scons python=2.5 binary_dist
and then I would modify the sconstruct script to do the rest. Is
there a reason you have to specify the interpreter as being explicitly
python2.X on the first line of the 'relax' file? If you are a
developer, have multiple python versions installed, and would like to
test this, this makes sense (but can be used by typing "python2.5
relax --test-suite"). For a normal user though, I would assume that
there would be one version installed and this would be what you get
when you type 'python'.
Cheers,
Edward
2009/8/18 Jack Howarth <howarth@xxxxxxxxxxxxxxxx>:
Edward,
Attached is the patch as applied to the tagged svn of relax 1.3.4
(I couldn't find the trunk svn). My other patches are specific to
packaging relax in fink. They concern the relocation of the relax
directories and suppression of the automatic compilation of the
python scripts (which for packages should be done in a post install
script).
Jack
On Tue, Aug 18, 2009 at 11:00:51AM +0200, Edward d'Auvergne wrote:
Hi,
That's awesome! Could you make a patch file that I can apply it (and
credit you for it)? For this, just type:
$ svn diff > some_patch_file_name
and then send the resultant file as a response to this. It should
only contain this line change, therefore you need the sconstruct file
to have only this change. Your other changes need to be incorporated
in another patch, and probably discussed as well. I would like to
make everything automatic for you, but we also need to take into
account that a user may have python and relax installed outside of the
Fink system (i.e. the repository 1.3 branch and some python version
installed as a user in the home directory or desktop).
Cheers,
Edward
Re: compiling 1.3.4 against python 2.6?