Re: [A-A-P-develop] Fake sources, one last time

[Adriaan de G. <ad...@cs...>]

On Friday 01 August 2003 10:35, Wichert Akkerman wrote:
> Previously Adriaan de Groot wrote:
> > I've finally figured out what I mean by "fake sources". All along Bram
> > has pointed out that _SOURCES should contain ultimate source files. Which
> > is fine as long as the structure of the source file name foo.ext is the
> > same as the name of the object file foo.$OBJSUF. If it's not - and in my
> > case, I'd like to generate foo_skel.o from foo.h - _then_ I call foo.h a
> > "fake source".
>
> That's highly confusing naming, since it is quite common to generate
> files with completely different names (lex, bison, docbook, etc.).
> 'fake' suggests the source does not really exist, which is not the case
> here.

Ugh, I meant "foo.skel".

Let me explain in detail what's going on here. There is a file foo.h, which is 
a plain C++ header file. It happens to contain some additional information, 
and the whole file can be translated into foo.kidl. foo.kidl can, in turn, be 
translated into foo_skel.cc, foo_stub.cc, foo_stub.h, and foo_signals.cc. Any 
given application needs only either the _skel, the _stub, or the _signals 
files, and needs to link against the _skel.o, _stub.o, or _signals.o. 

It's common practice in KDE Makefile.am's - but only because of our hacked up 
build tools - to list foo.skel in a list of sources when you want the 
application to link against foo_skel.o, etc.

I want to achieve pretty much the same thing in an AAP recipe. If worse comes 
to worse, I can always s/.skel/_skel.cc/ in the sources lines, and add a 
:rule to generate them, but I'd prefer to try for as little change to the 
Makefile.am as possible, in order to keep the barriers to possible adoption 
low.

-- 
pub  1024D/FEA2A3FE 2002-06-18 Adriaan de Groot <gr...@kd...>
     Key fingerprint = 934E 31AA 80A7 723F 54F9  50ED 76AC EE01 FEA2 A3FE