PEP 284, Integer for-loops (fwd)

[James_Althoff at i2.com]

[David Mertz]
|"Gnosis XML Utilities" has 3100 lines (written largely by me,
|with various contributors).
|There are *4* uses of [x]range() in 'for' loops.  One is a test case
|that just uses 'range(500)' to do something "a bunch of times".

[James_Althoff at i2.com]
|I just did the same for a small section of our code.  My quick little scan
|turned up 293 uses of [x]range.
|Our mileage might be different from others because our use of Jython is
for
|large-scale, shipping, production-level, commercial applications rather
|than for small, throw-away scripts or core-language libraries.  Also (as
|mentioned) we use Jython and have to deal with tables

[David Mertz]
|My utility package is certainly not "throw-away scripts".  And while
|it is generally tested with CPython, there's no reason it shouldn't
|work equally well with Jython.  So some of that difference isn't
|really germane.

You are right.  There are more categories and possible factors.


|The issue about iterating through data tables might be the key.

I think so too.  What is key in particular is that the table components
were designed outside of Python and independent of Python's standard
conventions.


|But I think even then, my own style would generally not use range()
|iteration.
|
|For example, one might write (very simplified):
|
|    rows = dbquery(sql)
|    for i in range(len(rows)):
|        print i, rows[i]
|
|But I would probably approach this more like:
|
|    i = 0
|    for row in dbquery(sql):
|        print i, row
|        i += 1
|
|It feels cleaner to me to iterate through the actual collection,
|rather than numerically over its length... even if I am then interested
|keeping track of the number.

That is nice alternative *when* you can say
    row in table

But if your protocol is
    def getRowCount(): pass
    def getColumnCount(): pass
    def getValueAt(rowIndex,columnIndex): pass
    def setValueAt(rowIndex,columnIndex): pass
then that option is not available.

|If one fetches rows incrementally, I would
|still probably go for a 'while 1:' style.

That is an alternative.  It seems unfortunate to me that one would have to
drop back to a "while 1:" style when dealing with something so common and
fundamental as indices.  But that's just my opinion.

Jim Althoff
Senior Vice President
i2 Technologies, Inc.