-L-ldl needed on Linux

[Unknown W. Brackets]

Yes, I'd think it's Derelict's fault personally.  On Windows, 
LoadLibraryA is (as far as I know) in the same library as a bunch of 
other things you're going to need.

On Linux, it's separated into its own thing.  It's potentially possible 
you might want to write a Linux executable that does not require libdl, 
specifically for a system which might not have it.

That said, it seems fairly unlikely.  Surely you can change the dmd.conf 
file to automatically pass -L-ldl to dmd, solving your problem?

-[Unknown]


Bill Baxter wrote:
> Unknown W. Brackets wrote:
>> Isn't this what the "lib" pragma is for, if you don't want to have to 
>> pass it?
> 
> I suppose so (or "link" pragma using dsss/bud).  So you're saying it 
> could be considered "Derelict's fault" for not providing that pragma.
> 
> On Windows, the equivalent libs (containing 
> LoadLibraryA/FreeLibrary/GetProcAddress) seem to be linked by default, 
> by someone in the toolchain.  DMD or DSSS, I'm not sure which.
> 
> So it seems reasonable that it should be there by default under Linux, too.
> 
> 
>> -[Unknown]
>>
>>
>> Bill Baxter wrote:
>>> Just wondering why I have to give dsss the -L-ldl flag manually when 
>>> I compile a Derelict-using program under Linux.
>>>
>>> Seems like it shouldn't be neecessary for me to have to specify that. 
>>> But who's to blame?
>>> * Is it DMD's fault for not including in the default link libs?
>>> * Is it DSSS's fault for not picking up that it's needed (or for not 
>>> puttin it in the default list of flags?)
>>> * Is it Derelict's fault for somehow not telling DSSS that it wants 
>>> to use libdl on Linux?
>>>
>>> Any of those seem possible.  So anyone know the answer?
>>>
>>> Or is it just my fault for thinking I shouldn't have to stick a 
>>> version(linux) block in my dsss.conf file?
>>>
>>> --bb