[Mono-devel-list] Bride of GAC

[John Luke]

Mike Kestner wrote:

>On Tue, 2004-05-11 at 16:26, Todd Berman wrote:
>
>  
>
>>>I read everything posted on this subject (GAC) I still don't get why 
>>>fully qualified paths to the assemblies are needed.  
>>>      
>>>
>>The best explanation I got from Gert was because of how nant works.
>>
>>However, the resolving from the GAC is fairly broken in general.
>>    
>>
>
>Just to be clear.  This has nothing to do with GAC runtime resolution. 
>This is a "compile-time" requirement. 
>  
>
Yes, I know that.

>Miguel's requirement is to have the assemblies only located on disk in
>one distinct location with no links or copies. We are putting the
>assemblies into the GAC so that they can be found easily at runtime.
>
>  
>
Yes, I agree with that.

>In order to compile against the assemblies, though, mcs has to be able
>to find them based on a provided -r.  We will be providing .pc files
>with references like:
>
>Libs: -r
>/usr/lib/mono/gac/glib-sharp/1.0.0.0_someobnoxiouskey/glib-sharp.dll
>etc...
>  
>
This is only necessary if compile time requires the full path, which is 
what I am questioning, and it very well may be necesary.  Otherwise the 
.pc files could just contain '-r:gtk-sharp' like they used to before 
0.91.   It just seems like they way it's going now works worse than 
before, all based on this one thing (requiring full paths to references 
to get the version), which may or may not be solvable in other ways with 
less adverse effects.