Re: pragma Linker_Options (was: JEWL on XP?)

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

"Georg Bauhaus" <rm.dash-bauhaus@futureapps.de> wrote in message 
news:491807ee$0$31337$9b4e6d93@newsspool4.arcor-online.net...
> Jeffrey R. Carter schrieb:
>
>> Depending on your compiler, you may need to convert the single "pragma
>> Linker_Options" into multiple, one for each item in the single one.
>
> So, pragma Linker_Options can be troublesome in case one
> cannot change the Ada source files.  (Because whenever
> a compilation system interprets the string_expression of the
> pragma and the linker does not understand the string,
> compilation fails.) Wouldn't it be better, in general,
> to avoid system dependent pragma Linker_options in source
> files?

Well,  maybe, but then you potentially lose version control on the options. 
And you might not have control over the building environment, but still need 
the options given.

For instance, in Claw, we have a package with separate bodies for each 
compiler. Some of those bodies contain one or more Linker_Option pragmas. If 
they're not given, Claw won't work (usually with a completely mysterious 
error message). We don't provide build scripts with Claw (the various 
compilers have fine mechanisms for that, we don't need to duplicate that). 
So pragma Linker_Options is the only way to ensure that the correct options 
are used.

I'd expect issues like these to arise for any "binding" packages that are 
distributed separately (assuming that you need facilities beyond those 
provided with the core OS). Similar reasons arise as to why you might want 
to use configuration pragmas rather than compiler options to suppress checks 
and the like.

                                  Randy.