Scalability Research (Was Re: Python suitability)

[Patrick Phalen]

[Darrell, on Mon, 27 Dec 1999]

:: The large scale idioms for C++ are well known. I'd love to read  "Large
:: Scale Python Software Design", when someone writes it.
:: 
:: Component interfaces like COM or CORBA make big stuff work and Zope offers
:: some such.  Although I wonder if they have too many good ideas in one place.

The subject of Python suitabililty for large projects seems to come up
more frequently of late.

Prior to Developer's Day 1998, Jim Fulton floated a pre-proposal for
Python "interfaces" (aka Scarecrow), which stimulated heavy initial
discussion on the types-sig. Why the types-sig? Probably simply because
that's where the Grand Masters were hanging out.

In any case, earlier this month, GvR slapped every body around on that
sig for not posting and threatened to dissolve it. Tim <winkingly>
complained that Guido was actually at fault, because, like Joyce's God,
he was off paring his fingernails and not participating in and
validating the discussions.

Jim Fulton (rightly, IMO) pointed out that momentum had tended to waver,
due to the tripartite nature of the threads:

* Interfaces
* Classes vs types
* Static typing

Three big subjects *is* perhaps a bit much for one poor little sig to
carry. Then again, perhaps the threads are simply migrating to c.l.p.

Nevertheless, the Eighth International Python Conference is coming up.
Looking at the schedule, including Developer's Day, I can find no
provision for *reviewing* progress made since the  7th Developer's Day
on interfaces, (nor classes vs types or static typing). Shouldn't a time
slot be made for this sort of progress report?

Quoting the back cover of _Learning Python_:

"Python is an object-oriented language for writing standalone programs,
quick scripts, and prototypes of complex applications."

This is, undoubtedly, a succinct, accurate depiction of Guido's
original intentions for the language. The problem is that he designed
such an attractive little prototyping language that an increasing
number of people are calling on it to do considerably more.

In his latest Regular Expressions, Cameron Laird predicts that Zope and
HP's e-speak will be killer apps for Python and that Python may achieve
critical mass next year. Sounds right to me.

The question then is, shouldn't IPC8 be a time to address these issues
and summarize them for us plebes?

Managing large-scale development isn't easy in any language and much
research effort goes into evolving new ways of grappling with the
problems. I wouldn't like to think that thinking and  research into
making Python safe for the large-scale project world was losing
momentum.