J2 proposal: keyword

[Michael Sparks]

On 22 Aug 2004, Paul Rubin wrote:

> "Robert Brewer" <fumanchu at amor.org> writes:
> > -Candidates for keywords have fallen into two or three camps, and
> > emphasize different aspects of decorators:
> >
> > -Declarative: declare, predef, moddef
> > -Transformative: transform, wrap, modify, mutate
> > -Attributive/Annotative: amend, using, having
> > -Directive: pragma, signify
> > -Associative: helper, qualify, qual, meta
> > -Cross-cutting: imbue, endow, bestow, embellish, extend, accum, glom,
> > confer
> > -Prepositions/Adverbs: using, through, per, via, by
>
> Not sure where these would go: prologue, preamble, preface
>
> How about something like "def_using"?

How is def_using different from "using ... def ..." ? ;-)

FWIW, I've spent a significant chunk of today after getting the patch for
J2 working looking at what keywords clash and what doesn't.

   * decorate, transform and declare have clashes with several easy to
     find projects. This includes commonly used things like mailman &
     pyrex.

   * "using" appears to be used only by python itself in webbrowser.py in
     the get function as a single named parameter, however which I can't
     find any project using it.

     I discovered this after replacing "decorate" with "using" and
     rerunning the test suite... Incidentally the syntax also allows
     code like this:

     using:
         staticmethod synchronised memoise deprecated
     def: foo(bar):
         "This function foos bars all day long"
         pass

     Personally, I think that's quite nice. I've not posted the updated
     patch yet because I'm looking at the short/single line version.

Projects I've grepped (and looked through the results) through include:
   wxPython, libxml2, SWIG, PyXML, ZODB4, distutils (as a standalone)
   egenix-mx, Redfoot RDF tools, PyLucene, dbxml
   libxslt, jabber-py, PyEgads, PyChecker, soappy
   m2crypto, epydoc, 4suite, Chandler

At some point Zope and Twisted need checking as well. (I'd like to check
most things, but those 2 strike me as important checks)

On a more general note, I think when suggesting a new keyword how widely
used a keyword is should be taken into account, which really needs to take
into account how widely used projects are that use the keyword.

At the end of the day, hopefully this helps Guido and everyone make a
decision, and helps build consensus either in favour of keyword based
(and a keyword), or towards the current @pie syntax.

Regards,


Michael.