Soapboxing. (was: re_comp and re_exec problem solved)
amolitor at eagle.wesleyan.edu
amolitor at eagle.wesleyan.edu
Fri Feb 22 07:38:23 AEST 1991
In article <1991Feb20.174215.27365 at csn.org>, cdash at mumm.colorado.edu (Charles Shub) writes:
>
> [some stuff in which it becomes clear that a regexp library is being
> used for globbing]
>
> charlie shub cdash at boulder.Colorado.EDU -or- ..!{ucar|nbires}!boulder!cdash
> or even cdash at colospgs (BITNET) -or- (719) 593-3492
I am unable to resist the temptation to soapbox. This is probably the
wrong place for this, but probably the right audience. I apologise in advance.
Skip over this post now. Last chance...
Why would one layer *more* code atop a regexp library for no reason
but to lobotomize it? Of course, it's because some user (or some boss) will
get upset if you try to force regexps down his or her throat. It is a Bad
Thing that software design is driven, seemingly more and more so, by
factors other than 'the *right* way to do it.'
It is fast becoming routine for a standard to appear, for everyone
to pretty much agree that it's a wretched standard, nigh-well impossible to
comply with, and for teams of programmers to be immediately dispatched to
write compliant software. It has long been routine for someone to get what
seems at the time to be a great idea, and to hack this great idea into some
existing code without, apparently, much thought and no concern at all for
pre-hack structure of the software -- the only thing that matters is 'it's
a great idea' or 'so and so wants it'.
The result of letting this sort of external factor control
software design and modification is all too often huge, slow products that
crash a lot, and don't *actually* comply with any standards at all. The
users, though, are happy -- the new product by-god looks exactly like the
software it replaces!
I submit that the world would be a better place if people designing
and writing software were to, more often, simply 'do it right'. This is
presumably obvious? There are no shortagle of examples of software done
wrong. I'll avoid naming names (a feeble attempt to dodge flames :-) but
for software done right, hmm, think white book. Slim paperback volume.
Of course, the external issues of continued salary and so forth will
inevitably rear their ugly heads -- I presume I high percentage of working
programmers would like to do it right, but aren't allowed to.
Andrew
amolitor at eagle.wesleyan.edu
{uunet, rutgers, generic Internet host}!eagle.wesleyan.edu!amolitor
More information about the Comp.unix.internals
mailing list