Re: encapsulating directory operations

"Lawrence D'Oliveiro" <ldo@nz.invalid> wrote in message
news:100jhh4$2lgt3$3@dont-email.me...

On Tue, 20 May 2025 22:26:57 -0400, James Kuyper wrote:
>

There is a practical barrier - while C was developed in the context of
Unix, which had a lot of influence on the design of C, that design has
also always been motivated by a desire to, among other things, be as
widely portable as possible.

>
Furthermore, back in that day, some OSes (like MS-DOS) were pretty limited
in their functionality, without even multi-tasking capability.

And note that I sitll expect the proposed C90+ to support MSDOS.

That situation has improved somewhat. Nowadays the world divides into
POSIX-like versus non-POSIX-like. With the latter trying to become more
POSIX-like over time.

I'm not interested in restricting myself to "nowadays".

That doesn't mean I want to add a "far" keyword though.
Not that I would rule it out either.

And MSDOS 4.0 - one of the systems I am interested in -
is not "trying to become POSIX-like".

The C library I plonk on it can be updated though.

The most different structure that I'm personally familiar with is
VMS, where the closest equivalent to a Unix directory was versioned.
If you specify a directory with a version number, you get that
version of that directory (if it exists); if you don't specify a
version, by default you get the latest version.

>
Only regular files were versioned; directories were files, but they were
specially managed by the OS, and they had to have the .DIR suffix and the
version number was always 1.

Sounds perfectly capable of responding to the directory
routines I postulated. ie VMS support could be added
to folder.c.

BFN. Paul.