[Info-vax] A DCL wish list of sorts...

[Craig A. Berry]

On 3/22/19 6:29 AM, pcanagnostopoulos at gmail.com wrote:
> Fortunately, people designing programming languages are now stricter
> in their specification of the lexical and syntactic elements. Well,
> except for Perl.

Perl was mostly designed 25-30 years ago and has gotten a lot stricter
and provided a lot more warnings and strictness capabilities in recent
years.  It is difficult to change because people in many cases have come
to depend on emergent behavior and scream bloody murder when any of it
gets changed in an incompatible way.  A good example was that it was at
one time possible (quite by accident) to include a control character in
a variable name, but the documentation warned against doing so.
Eventually, that was turned into a syntax error.  One person who had
written one evil line of code with a control character in a variable
name raised quite a ruckus in the form of long, obscenity-laced
diatribes to the developer list accusing us of all sorts of perfidy for
breaking documented behavior (even though what the documentation said
was "don't do that").  People also have a habit of reading the code and
using APIs that are not marked public and then raising a fuss when those
APIs get removed or refactored.

Such things are not unknown in the VMS world.  7.3-2 made some changes
to RMS internals that broke some of the more hackish RMS code people had
written.  And people used to depend on monotonically-increasing process
IDs, etc.