[lsb-discuss] Looking for comments on Perl, Python (and why not TCL, Ruby, PHP, etc.)

R P Herrold herrold at owlriver.com
Tue Jun 5 19:13:13 PDT 2007


(This will break threading, -- sorry -- but the piece from one 
of the Red Hat mailing lists needs to be seen and considered. 
As I see it there is just not a buy-in and essentially zero 
interest in coding to a stable Python subset/featureset over 
time.  This is in part, driven by the Pythonic culture's 
seeming willingness to break back API's because it seemed like 
a good idea at the time  ;)   perl has more mass, and so seems 
reasonably well ossified in the 5 series, and with a 
reasonably tractable CPAN add on module culture
  -  Russ)

Wichmann, Mats D said:

> I don't want to seem dense, but I'd really like to try to 
> raise this question again so we're sure everybody is on the 
> same page.

> We get a fair bit of comment that LSB needs to specify Perl 
> and Python, and occasionally (MUCH less frequently) other 
> interpreted languages such as TCL, Ruby, PHP.  I'm 
> intentionally leaving Java out since there are some 
> different dynamics involved in that question; if it were 
> included it would unquestionably be the highest demand in 
> this general category.

I tossed an enumeration of RHEL perl, python versions at:
 	http://lists.freestandards.org/pipermail/lsb-discuss/2007-May/003875.html
as we do not get a @redhat.com representative on the 
conference calls with any consistency.

Mats continued:

> It's becoming increasingly clear that this level of detail 
> will not happen for Perl and Python, definitely not in the 
> short term and probably never.

After reading the following, is there anyone left who cannot 
conclude as to Python that 'never' may actually be optimistic  	;(

Really if an ISV wants perl, that is probably 'doable' -- 
Python?  they'd better be willing to carry a local copy with 
the features they need in their installer, if they intend to 
use it.

-- Russ Herrold

quoted matter follows:
---------- Forwarded message ----------
Date: Tue, 05 Jun 2007 17:20:51 -0400
From: Jeremy Katz <katzj at redhat.com>
Reply-To: Development discussions related to Fedora Core
     <fedora-devel-list at redhat.com>
Subject: fedora-d-rh] Re: Feature idea: package an installer image as a grub
     entry  before F8.

On Tue, 2007-06-05 at 17:10 -0400, Colin Walters wrote:
> On Tue, 2007-06-05 at 16:38 -0400, Jeremy Katz wrote:
>> While we're at it, can we disallow GUI apps from using new GTK
>> features? ;-)  More seriously, adding restrictions like this doesn't
>> really help us to improve things for our userbase because we either end
>> up having to reimplement things that appear for the language or do weird
>> hoop-jumping.
>
> Are we just talking about Python or are there more problematic things?
> Would it really be so bad to be fixed at Python 2.4?  What's so hard to
> reimplement from Python 2.5 features?  From
> http://www.python.org/download/releases/2.5/highlights/
> I don't see much.

We actually already have a fair bit of pain to keep yum 3.2 working with
python 2.4 and 2.5.  Things like the integration of cElementTree and
pysqlite both led to API changes that we're carrying compatibility cruft
for.  The former isn't too bad.  The latter regularly makes me want to
stab myself repeatedly in the eyes.  We're not taking advantage of the
new try/except/finally bits which actually hurts quite a bit -- and just
trying to emulate or reimplement it _doesn't_ work nearly as well.

And even if there wasn't _anything_ in the python 2.4 -> 2.5 transition,
that doesn't mean that there hasn't been in the past and won't be in the
future.  Why would we want to tie the hands of a project like that?  And
don't forget -- it's not necessarily just python... it could be anything
else that we depend on.  rpm-python/rpm being high on the list of
probables.

Jeremy




More information about the lsb-discuss mailing list