[ Date Index ][
Thread Index ]
[ <= Previous by date /
thread ]
[ Next by date /
thread => ]
On Mon, Jan 28, 2002 at 05:24:08PM +0000, Adrian Midgley wrote:
One of the points of it would be that it is quickly put together from available pieces (high quality well-tested modules with known behaviour I mean), tied together with a very little code written to perform a specific task.
Try not to give too much false hope to the idea that simple software,
such as this, would be better done by writing abstract modules and
then putting them together with a little perl glue.
In my experience, "lego programming", as I call it has the following,
unfortunate side effects:
1) if the module is written:
a) Modules are often "complete", in as much as they do everything
in context, not just what you want. This can cause bloat.
b) Modules can sometimes do things in ways you don't want, under the
banner of "correctness".
b) One cannot rely on the qualty of the code written and it is often
easier to write what you want than debug someone elses code.
2) if the module has not been written:
a) Too much time is put into writing a "complete" module which can
be used for other purposes.
b) Interfaces are written to allow for generic use, as opposed to
what you might want, in context.
Theo's mod_age is superb; I like it a lot. I'm not saying it's not
useful, but it may be folley to say "right, we've got this module, now
we must use it", when a few lines of application specific code would
have done the trick. It may be the case, in fact, that it is the best
solution, but there is no reason to make that kind of descision at
this stage.
I remember an example of a freidn who asked a question about
dates. They wanted to format a date in a specific way. There were
people raving on about this module and that, but it came down to one
line of perl to do the trick. CPAN is great, I use bits of it, but
using modules for the sake of it is silly.
Better, I say, to write the thing, and then, if time permits, rips out
sections which fall into obvious modules. Then, if time permits, fancy
up those modules for reuse.
Reuse should not be the primary goal of software development, solving
the problem in hand should be.
The story being told to much of the health service, including general practices who have run terminals on SCO Unix for years, is that email can only be done on Windows, that Word documents cannot be read without Windows, and that all of this is easy to use and secure against supplier variation and intruders.
Eww!
One has to wonder whether the NHS planning and management performance is distinguished by its level of effectiveness in IT, or if it manages a uniform effectiveness across a wide range of management and business activity.
Who dares to speculate? Steve -- The Mailing List for the Devon & Cornwall LUG Mail majordomo@xxxxxxxxxxxx with "unsubscribe list" in the message body to unsubscribe.