El Tue, 29-09-2009 a las 12:36 +0000, Aleksey Lim escribió:
pygame and numpy are parts of Sugar Platform[1], at least for 0.84, so, the right question is should these pakcages be a part of SP-0.86 I guess +1 for both, since we have honey activities that are depend on these packages.
In Sugar shell 0.82, the only thing that numpy was providing was a bit matrix. In 0.82, we replaced it with faster and lighter C code in sugar-base, and it was a big performance win.
I suspect the #1 usecase for numpy is to compensate for lack of good array support in Python.
Questions:
1) are there lighter-weight alternatives for the most popular uses of numpy?
1) how many of the existing activities actually depend on numpy?
2) would it be hard to remove this dependency from them?
3) Should we define a policy for deprecating components of the Sugar Platform in new revisions? All evolving standards need to find a balance between new features with old feature removal to avoid unbounded bloat.
4) Even if numpy is going to stay around for the Sugar Platform, could we remove it from Pippy and other core activities to save resources and allow shipping lighter weight live distros?
On Tue, Sep 29, 2009 at 9:10 PM, Bernie Innocenti bernie@codewiz.org wrote:
El Tue, 29-09-2009 a las 12:36 +0000, Aleksey Lim escribió:
pygame and numpy are parts of Sugar Platform[1], at least for 0.84, so, the right question is should these pakcages be a part of SP-0.86 I guess +1 for both, since we have honey activities that are depend on these packages.
In Sugar shell 0.82, the only thing that numpy was providing was a bit matrix. In 0.82, we replaced it with faster and lighter C code in sugar-base, and it was a big performance win.
I suspect the #1 usecase for numpy is to compensate for lack of good array support in Python.
Questions:
- are there lighter-weight alternatives for the most popular uses of
numpy?
- how many of the existing activities actually depend on numpy?
A quick repo query gives me pippy and sugar-speak in the fedora repo.
Peter
On Tue, Sep 29, 2009 at 4:10 PM, Bernie Innocenti bernie@codewiz.org wrote:
El Tue, 29-09-2009 a las 12:36 +0000, Aleksey Lim escribió:
I suspect the #1 usecase for numpy is to compensate for lack of good array support in Python.
Questions:
- are there lighter-weight alternatives for the most popular uses of
numpy?
How about numarray?
$ rpm -q --requires python-numarray /bin/sh /usr/bin/env libc.so.6()(64bit) libc.so.6(GLIBC_2.2.5)(64bit) libc.so.6(GLIBC_2.3)(64bit) libc.so.6(GLIBC_2.3.4)(64bit) libc.so.6(GLIBC_2.4)(64bit) libm.so.6()(64bit) libm.so.6(GLIBC_2.2.5)(64bit) libpthread.so.0()(64bit) libpython2.6.so.1.0()(64bit) python(abi) = 2.6 rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PartialHardlinkSets) <= 4.0.4-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1 rpmlib(VersionedDependencies) <= 3.0.3-1 rtld(GNU_HASH) rpmlib(PayloadIsXz) <= 5.2-1
- how many of the existing activities actually depend on numpy?
Peter answered this
- would it be hard to remove this dependency from them?
I seem to recall several Python apps moving from numpy to numarray in the past, so it should be doable. If we want to do this, we should probably create a tracker bug in Bugzilla.
- Should we define a policy for deprecating components of the Sugar
Platform in new revisions? All evolving standards need to find a balance between new features with old feature removal to avoid unbounded bloat.
That sounds reasonable. Like "new activities using numpy will not be accepted"? If the porting work from numpy to numarray is documented, this should be linked to from the policy as well, so that people can adapt their activities.
- Even if numpy is going to stay around for the Sugar Platform, could
we remove it from Pippy and other core activities to save resources and allow shipping lighter weight live distros?
Probably a good thing to do, yes.
Regards,
On 09/29/2009 02:35 PM, Michel Alexandre Salim wrote:
On Tue, Sep 29, 2009 at 4:10 PM, Bernie Innocentibernie@codewiz.org wrote:
El Tue, 29-09-2009 a las 12:36 +0000, Aleksey Lim escribió: Questions:
- are there lighter-weight alternatives for the most popular uses of
numpy?
How about numarray?
http://www.stsci.edu/resources/software_hardware/numarray numarray is being phased out and replaced by numpy. STScI has migrated all of its software to use numpy, and the current release of stsci_python and STSDAS/TABLES uses numpy in place of numarray.