[PD-dev] loaderHEX bug report

[Hans-Christoph Steiner]

On Nov 22, 2005, at 10:43 AM, IOhannes m zmoelnig wrote:

> Hans-Christoph Steiner wrote:
>
>>> in build/mtx_0x2a.c the wrong setup-function is defined.
>> Ah, thanks... it was a long day.  Did you get the other bug report?    
>> That one is a bug in the patch... really!  ;)
>
> yes, i  know though i haven't made any steps to commit a fixed patch  
> yet...
>
>> I have a question, why did you reverse the order of the original   
>> stantard setup function?
>> i.e. original = mtx_+_setup()    altname =  setup_mtx_0x2a()
>> I think that just makes things confusing and causes bugs, well, like   
>> the one we just talked about.
>
> yes it is confusing but:
> i want to have a different setup-function for the alternative names;  
> otherwise it would forbid to make an object that is really called  
> "mtx_0x2a". on 2nd thought, this is _almost_ impossible now too,  
> because it is hard (though possible) to have 2 mtx_0x2a.pd_linux lying  
> around (one for the [mtx_*] and one for [mtx_0x2a])
>
> what is more important is, that you cannot start function names with  
> numbers (at least in C)
> so the [<<<] object (which would expand to "0x3c0x3c0x3c") cannot have  
> a valid setup-function "0x3c0x3c0x3c_setup()"; however  
> "setup_0x3c0x3c0x3c()" is fine.

What about alt_0x3c0x3c0x3c_setup()?  That would highlight that its an  
alternate name, rather than just having an seemingly arbitrary reversal  
of the "setup" placement.  And the function is called  
sys_load_lib_alt().  That makes more sense in terms of the API.

> so i really think, that there should be 2 separate names (if we agree,  
> that it would be a bad idea to rename the standard setupfun-name to  
> setup_***()

Yes, that is more trouble than its worth.

.hc

________________________________________________________________________ 
____

                   ¡El pueblo unido jamás será vencido!