Re: Community Input for the Maintenance and Revision of the Ada Programming Language

["G.B." <bauhaus@notmyhomepage.invalid>]

On 03.10.17 12:11, Dmitry A. Kazakov wrote:
>  It is not very likely you would actually use strings larger than 32K.

When a program reads a text file, of size > 32K, into some string
("into memory"), this can make text processing a lot easier. Just
think about implementing an abstraction that allows scanning the
entire text in case there is no string large enough to contain it.

Writing a parser seems less likely and more demanding than feeding
a 32+K string to some text processing routine (counting, searching,
substituting, ...). There are fewer corner cases to consider, the
program does not need state for remembering the previous line's
final words, and so forth.

> Now if size/bound becomes a type or object property that makes some programs illegal.
> 
> Arguably that could be desirable if the bound comes from the problem space. However in most cases it is just more or less arbitrary design choice which will hit you hard later if you were too generous.
> 

Indeed, suppose designating parts of strings, such as elements or
slices, were available through some types more targeted than mere
Positive. This choice directly leads to more flexibility, too.

For illustration, making Swift 2, 3, and now 4 has included much work
on overhauling NSString, so that now, in current Swift, it has
become String, and Substring, together with associated types for
iteration, slicing, and so forth.


String is the aftermath of our own bias when commencing the
work of formalization by analyzing our writing. It permeates
all programming to an extent that may be in the way of seeing
proper abstractions. It's close to making

    array (Positive range <>)

be so ubiquitous that it appears in the public interface
of just about any private type. And the above one doesn't,
does it?