Re: a new language, designed for safety !

[Robert A Duff <bobduff@shell01.TheWorld.com>]

Jeffrey Carter <spam.jrcarter.not@spam.not.acm.org> writes:

> On 06/16/2014 02:53 PM, Robert A Duff wrote:
>>
>> It's also unfortunate that Vectors are inherently slow.  

To be clear:  I meant that Ada.Containers.Vectors as defined by the Ada
RM are inherently slow.  Not that all possible vector-like
(i.e. growable-array implementations) are inherently slow.

>>...I wrote a big
>> program recently that made heavy use of Vectors.  The Ada 2012 features
>> were essential for readability -- Vectors would have been unusable
>> otherwise.  But Vectors was way too slow.  So I wrote my own vectors
>> package, using the same 2012 features, and then it was fine -- both
>> readability-wise and efficiency-wise.
>
> If Vectors are inherently slow, how were you able to write one that wasn't?

By writing one that has different semantics -- i.e. does not obey all
the Ada RM rules for Ada.Containers.Vectors.

For example, it has the operations I happened to need (like Append), but
lacks the ones I did not (like Insert).

For example, it supports easy and efficient conversions
Vector <--> array.

For yet another example, it lacks the "tampering" business required by
the RM.  I didn't need all sorts of Cursors pointing into the middle of
Vectors, so I didn't need those run-time checks.

The new Ada 2012 features made my vectors usable from a readability
point of view, just as they have for Ada.Containers.Vectors.

- Bob