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

["Vasiliy Molostov" <molostoff@gmail.com>]

Georg Bauhaus <rm.dash-bauhaus@futureapps.de> писал(а) в своём письме Mon,  
20 Aug 2012 20:22:50 +0400:

> On 20.08.12 16:15, Dmitry A. Kazakov wrote:
>
>
> Related Example that has been part of this discussion,
> and concern compiler configuration files:
>
> A software program to be built comprises, among other things,
> four projects.
> Three of them are written in Ada, one is written in C++.
> Two use GNAT, one uses Janus/Ada, and the fourth uses Visual C++.
> Every now and then, the teams need to cooperate to produce one
> executable from the libraries that each team will have built.
>
> Let each team need to support three scenarios:
> Development, Integration/Testing, Production.
>
> Management says: "Switch to Integration/Testing this afternoon!"
>
> What then happens is that each team changes their tool chain's
> setup. Each team does so in a different way, so as to make the
> different compilers run as required for Integration/Testing.
> Heterogeneity.

Excellent example. I doubt that all four will follow some ruling as they  
have different tools and
there is probably no asis in VisualC++.

but the question still open for Janus or GNAT, since a presence of such  
xml access can make tools
like wsdl2aws, aws2wsdl, soap2ada, ada2restful, ada2java, ada2vhdl, etc  
useless, because any developer/tester (even non-ada) can dig xml and write
a generator (namely, xml transform) of any desired kind from a compilation  
result, in the same way it is generated from xmi (uml).

So as for example, this means that VC++ team can receive this compilation  
result in xml and generate from it everything they want.

A tester (even non-ada) can generate a set of tests from this input too  
(even into another language, say based on ajis), with little help from  
real ada developer. This means that Ada developer will think about Ada,  
and a tester would not. It is possible to automate documenting (extracting  
doc from source) without touching these sources, but just from compilation  
result (if it is in xml).this can drastically minimize effort and can  
provide a way to separate job in several pieces, where not well qualified  
personnel (non-ada) can dig these pieces for utility purposes by  
themselves.

But they all still need a compiler.