Re: Three simple questions

[Laurent Guerby <guerby@acm.org>]

kenner@lab.ultra.nyu.edu (Richard Kenner) writes:
> [...] Some would say this argues in favor of underscores not being
> significant either, so you couldn't have a disinct Roll_Control
> variable either, but "roll underscore control" sounds a lot smoother
> than saying "each cycle we set the capital r roll capital c control
> variable from the average of the lower-case r roll lower-case c
> control, lower-case r roll capital c control, and capital r roll
> lower-case c control variables"!

Nice example ;-).

As for underscores, if you put them consistently between words, you
don't even need to mention them while discussing your program. And
since Ada identifiers cannot begin with, have two consecutive or end
with an underscore, there's no problem at all.

Note that this issue transcend programming languages and goes up to
the more general human-computer interface field. Having case
sensitivity there is even worse and sometimes can be dangerous (source
of big mistakes once someone introduce XYZ and XYz in the system...).

The software I work on for my company (in Ada) has an API based on
names (object/attibute), and we follow and document the Ada rules for
the allowed set of names. Other software is written on top of usually
using another language (C, C++, Java, VB, you name it) by other
developpers. This layer often adds case sensitivity (ignoring our
documention rationale) and I remember a meeting with a manager who
asked the guyes who did the software in C++ using our API "are you
sure case sensitivity is a good idea". Needless to say, after a
meaningful silence the consensus was "no", but the software wasn't
changed because it would be too costly...

-- 
Laurent Guerby <guerby@acm.org>