[lsb-discuss] ISV Survey- Perl,Python
robert.schweikert at mathworks.com
Tue Jul 17 12:19:11 PDT 2007
Cool, I think this is getting more useful now. However, there is still a
catch we have to answer/resolve.
The spec says Python 2.4 or later. While I understand that the LSB
cannot force distributions to schlep Python 2.4 for all eternity and we
don't want to add Python 2.4, then deprecate it right away only to add
2.5 and repeat forever more, we need to somehow find a way to "freeze"
the behavior via the spec.
Right now the links for the modules point to the 2.5 documentation when
they probably should point to the 2.4 documentation. The way I interpret
what we are trying to do is as follows:
Distributions may supply any version of Python (2.4 or later).
Applications can depend on the 2.4 interfaces.
This means if in 2.6 one of the interfaces changes, app vendors still
can get the 2.4 behavior. Otherwise we just broke the LSB deprecation
policy (assuming 2.6 is available less than 6 years after 2.4 was
released). Therefore the documented interface behavior should be the 2.4
documentation and not the latest (currently 2.5) or we would change the
spec via link every time a new version of python is released.
IMHO we should clearly document this intend and we should also alert
users to the fact that the links point to what the LSB wants to include,
i.e. documents describing the 2.4 behavior.
For cPickle for example that would be
http://docs.python.org/lib/module-cPickle (this is the 2.5 version
and will be 2.6 in the future).
I know we do not want to duplicate the documentation but it might be
worth it to just host the Python 2.4 docs as a copy on the LF site to
assure that we don't end up with broken links if the Python folks decide
to change their site or worse yet end up pointing to the wrong interface
Stew Benedict wrote:
> On Tue, 17 Jul 2007, Robert Schweikert wrote:
>> The LSB module list should not contain any names which are not listed here:
> List trimmed, new spec doc:
> lsbappchk-python is updated to reflect the changes (in bzr), will go back
> over the runtime test.
Robert Schweikert MAY THE SOURCE BE WITH YOU
(robert.schweikert at mathworks.com) LINUX
The MathWorks Inc.
Phone : 508-647-2042
More information about the lsb-discuss