[
Lists Home |
Date Index |
Thread Index
]
On Mon, 05 May 2003 04:08:43 -0500, Eric Bohlman <ebohlman@earthlink.net>
wrote:
> I think the key here is that Saxon is "small enough" for one person to
> "get himself around it" completely, and the methodology you describe is
> entirely appropriate for such cases. XP, on the other hand, seems to me
> to be a methodology intended for programming projects that are "too big"
> to be entirely the purview of one person.
I had a very similar reaction to Michael Kay's response. But there's a
third option, "make everything small enough for one person to do
completely." That's more or less the idea behind modules which begat
objects which begat components which begat services -- "small pieces,
loosely joined."
The problem I see is that it is so very hard to design these small,
implementable, extremely useful components/services. Saxon clearly
qualifies, and but it a) requires someone of Kay's immense experience and
skill to do well; b) stands on the shoulders of James Clark and the others
who developed and refined XSLT; c) stands on the meta-shoulders of the
people who worked on DSSSL; d) stands and the meta-meta shoulders of the
people who developed Scheme .... and all this took a couple of decades to
sort itself out.
People who make a living doing more prosaic tasks of making less inspired
bits of software work together so that they can make Suzy Webmaster, Joe
Techwriter, and their Pointy Haired Boss more productive are the ones who
are more likely to benefit from an XP-ish approach.
|