On Fri, May 9, 2014 at 3:03 PM, ahi <[email protected]> wrote:
> On Fri, May 9, 2014 at 2:51 PM, Hin-Tak Leung
> <[email protected]> wrote:
>> I am not too concerned about python 2.7 vs python 3 - most of the world
>> is still operating at 2.7, but forward porting, one-off via the 2to3 script,
>> is relatively straight-forward; and backward compatibility, at least to the 
>> last
>> few minor releases of 2.7 or even 2.6, can be achived via "from __future__ 
>> import *".
>> It is possible to write python code that works identically under
>> 2.7 and 3. That's not really a concern - we'll cross that bridge when we
>> need to, when we have some code to discuss on.
>
> I have to say that I am more comfortable with python 2.7 but that's
> because I don't have any experience with python 3, but I am willing
> to use python 3 if its for the best.

Go with Python 2.7 IMHO, portability to 3 is not a problem and not an
issue until Python 3 becomes mainline in some (likely distant) future.

> On Fri, May 9, 2014 at 2:51 PM, Hin-Tak Leung
> <[email protected]> wrote:
>> The jython + existing java code ideas serves a few purposes - the python 
>> binding
>> should be feature-complete versus existing work; also, while it is nice
>> that Ahmed will continue after the program finishes, and/or others
>> will pick up the maintainence of such work, if it shares some
>> with a larger bulk there is a better chance of that happening -
>> otherwise it is just a few months of work gradually bit-rotting and
>> abandoned in a few years' time. I am okay
>> with it sharing some work with another language-binding project - otherwise
>> you get two sets of bugs and two sets of possibly incompatible
>> implementations to worry about; one of them gradually bit-rotting and go 
>> unmaintained.
>> I won't insist on it; it is just a possibility that should be explored.

The Java implementation can already be used in Jython as-is, so there
is no work to be done there?

On Fri, May 9, 2014 at 3:03 PM, ahi <[email protected]> wrote:
> I can definitely see the merits of this approach, creating a C library
> from the Java binding or from scratch could definitely benefit future
> projects. I am also ok with having a pure Python implementation from
> scratch.

I strongly think this would not be a good thing to work on bindings,
especially with Java.
These usually create portability issues and raise the bar for build
and deployment requirements for no good reason.
A native language implementation is IMHO best.

What makes sense as Nuno suggested would be to create a standard test
suite that is not language dependent.

-- 
Cordially
Philippe Ombredanne
_______________________________________________
Spdx-tech mailing list
[email protected]
https://lists.spdx.org/mailman/listinfo/spdx-tech

Reply via email to