[Info-vax] DCL's flaws (both scripting and UI)

[Bob Gezelter]

On Wednesday, January 21, 2015 at 5:23:19 PM UTC-5, JF Mezei wrote:
> Just a comment:
> 
> There have been suggestions about adding looping. (I use that as an
> example).
> 
> 
> One can already do looking in DCL, and except for ability to do a loop
> on a single line which you can do in bash, there isn't much functionlity
> that is prevented.
> 
> The focus should be on giving DCL functionlity that is missing or
> extremely hard/complex to do.
> 
> for instance: make it possible do do
> 
> $edit "Recette de gâteau au chocolat.txt"  instead of the extremely
> complex and non obvious escaping needed in DCL/VMS.
> 
> 
> Another suggestion: once VMS 1.0 is available on that architecture I
> can't name without making people angry, VSI should poll those customers
> who do migrate to that architecture on what features they are using (and
> this need upward compatibility).
> 
> This may allow VSI to make non-upward competible changes to VMS on that
> platform, knowing customers are not dependent on old behaviour and thus
> enable new features that way without being hampered by customers with
> legacy code who aren't gonna upgrade anyways.

JF,

Breaking existing behaviors is a non-starter. Even if you ask customers.

The simple fact is that almost all customers rely on a collection of products, which in turn use a variety of tools and toolsets. Breaking an established behavior is not considered a problem, UNTIL one of these foundational items fails to behave as expected. When ODS-5 was adopted, the community discovered that the case-folding behavior of ODS-2 was an unspoken presumption in more than a few command procedures, and the resulting bizarre behaviors caused many a telephone call.

Even small syntactic changes (e.g., extending the unambiguous abbreviation rule) can have significant impact. Many sites simply do not have the proficiency to dig through their historic (and vendor-supplied) DCL to resolve such problems.

Adding UNTIL/WHILE would not be too bad, but we would have to enable/disable the feature using something akin to F$VERIFY on a file-by-file basis (default would have to be DISABLED).

- Bob Gezelter, http://www.rlgsc.com