On 3/11/07, Guido van Rossum <[EMAIL PROTECTED]> wrote:
On 3/11/07, Andrew Clunis <[EMAIL PROTECTED]> wrote:
> Antoine van Gelder wrote:
> > Don Hopkins wrote:
> >> Of course it might just be more powerful and efficient to re-implement
> >> something like Klik-and-Play from scratch in Python, as a plug-in
> >> visual scripting component, which can be used to script a
> >> HyperCard-like gui environment, and games built on top of it like
> >> SimCity and Robot Odyssey.
> >
> > Anyone know the current status of being able to turn an AST tree emitted
> > by the Python compiler module back into code after it has been modified ?
>
> I've been looking into this for Develop activity, but it seems that path
> has never been explored. Best approach seems to be to write a
> pretty-printer to walk the tree.
I'm curious about the focus on ASTs that seems apparent in this
subthread (though I may easily be misreading between the lines :-).
I've always been more inclined to edit the text and re-parse from
there, as it puts the author in control of formatting, comments etc.,
and this is how most "real-world" environments work. (Not that that
necessarily makes it better, but neither is the opposite true.) Is
someone willing to write up a brief comparison between the two
approaches?
This isn't really a comparison, but briefly,
AST vs. source code
* constraining the data to be at least valid, syntactically correct
trees rather than ASCII chars in a row removes a major source of bugs.
* more amiable to direct manipulation via GUIs.
* can be elaborated w/o clutter vs. comments and annotations in text
which must be made to fit (function-annotations discussion on py3k
list)
* I18N
* compression.
* pre-processed for other tools, lint, translators, doc generators, verifiers
More important than AST vs. char[] is maintaining the ease of
inter-conversion of the two. The policy on python's grammar is a good
example of this principal.
Peace,
~Simon
_______________________________________________
Sugar mailing list
[email protected]
http://mailman.laptop.org/mailman/listinfo/sugar
