Re: Thanks Guys: Re: Predefined Packages

[Gene <gene.ressler@gmail.com>]

On Jun 15, 5:28 am, a...@anon.org (anon) wrote:
> Thanks for the info. I knew that Gnat had too many packages with in
> it structure.
>
>    And after finding a number of packages in GNAT that initially do not
> work.  An example is the "System.Partition_Interface" or files
> "s-parint.adb" and "s-parint.ads" which should allow one to read the
> partitioning tables but GNAT says this is a dummy package unless
> you download another of its many software packages that contains
> an replacement package. Why include this package in the first place
> just include it in the extra software package only.  So I decide to ask
> the question.


Hard to tell from what you wrote here.  I seem to be missing the first
part of the conversation.

Anyway, Partition_Interface is about Ada Annex E partitions in
distributed systems.  It would be pretty useless to have the
functional version if your system isn't configured with Annex E (i.e.
Glade).  Glade is available.  If you want to check it out, go get it.
Perhaps you're thinking this package has something to do with disk
partitions?  This isn't the case.

Other library stubs are there for similar reasons.

Having the functional code in the libarary would be pure bloat.
Having stubs can be useful.  For example, you can compile and run some
Annex E code on a machine without Annex E.

I agree that the existence of the stubs is not well-documented.  A
compiler warning would be nice.

Yes the GNAT package structure is fine-grained, but I diagree that the
system has "too many packages."  The packages are for the most part
nicely aligned with the objects or logical constructs they handle
(i.e. sections of the ALRM).  If you have a reasonable environment,
the number and size of packages ought not to be worth worrying about.