Re: Should Inline be private in the private part of a package spec?

["Randy Brukardt" <randy@rrsoftware.com>]

"Georg Bauhaus" <rm.dash-bauhaus@futureapps.de> wrote in message 
news:50211aed$0$6571$9b4e6d93@newsspool3.arcor-online.net...
...
> Premise 1: The reader is the only important person in Ada programming.
> Premise 2: The reader wishes to learn from a package spec how to write
> a call statement (essential knowledge: profile, contract, exceptions).

Premise 2, as you've stated it here, is clearly wrong. It would be much 
better stated as

The reader wishes to learn from a package spec whatever they need to know 
about using the package.

That might be how to write a call, but it also might be how to write an 
object declaration, a type extension, an access type declaration, or one of 
many other possibilities.

They also might want to know how to compile the program successfully, either 
because they want to compile the program by hand, or because they're some 
sort of tool. (Most of the Janus/Ada tools only read the context clause and 
visible part of packages; we don't want any privacy leakage and we 
definitely wanted to avoid reading large specifications.)

Point is, that it's pretty hard to tell want a reader wants. As Vasilly 
said, it's very annoying to have to look all over the place to find the 
information you need. Moreover, having a bunch of aspects at the end of a 
subprogram declaration has almost no effect on its readability, because you 
don't need to read past the "with" unless you care about the aspects. (This 
is just like not reading the private part if you don't need to know 
something about that.) So I see no advantage to the style of putting things 
in the private part, and there are lots of disadvantages (hidden 
dependencies, hard to find the information when it is needed, hard to tell 
when it is missing [because omissions are much harder to find than 
commissions], and much more likely to run into freezing problems).

                                 Randy.