[s1mp3-dev] C compliance was Re: [Swan API] STDIO module

[Legolas]

Bluechip ha scritto:

>>>>>I would also highlight the fact that the Swan API is *not* C compliant.
>>>>>
>>>>Right.
>>>>
>>>hy have you chosen to drop C compliancy?
>>>...It seems (to me) like a VERY weird choice.
>>>
>>Oh, I have chosen nothing. I just thought you were reporting a choice
>>done before I had joined the project
>>So the filesystem module needs a C compliance rewrital.
>>    
>>
>You asked my advice on this module the other day (off-group) - I 
>guessed that your choice to ignore that conversation was deliberate 
>rather than accidental.
>  
>
I just thought: "we are not going to be 100% C compliant because we 
probably can't".
I think we can write the OS-level C compliant functions without problems 
with this hardware, so I simply misunderstood your statement.

I suggest to add an introductive paragraph to 
http://www.s1mp3.org/wiki/index.php/Code_conventions to explicitly state 
that we will follow the C standard as best as we can; are you going to?

--
  Legolas

>BC
>
>
>  
>
>>>>>I would like to documentate the file system API as a "job" for 
>>>>>the API team.
>>>>>          
>>>>>
>>>>The FS API is defined by 'C' ...it is just a matter of implementing it.
>>>>fopen
>>>>fclose
>>>>fseek
>>>>fread
>>>>fwrite
>>>>        
>>>>
>>>You are right, the only real change I proposed was not to use a 
>>>null terminated string for the modes, but it is a minor change, indeed.
>>>      
>>>
>>NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
>>
>>ANSI C ...the saving is not worth breaking a 30 year-old standard for.
>>    
>>
>
>(ok) I really didn't think about it...
>
>  
>
>>>>there is no standardised way to access directories in 'C' ...I 
>>>>would suggest:
>>>>findfirst
>>>>findnext
>>>>as the simplest solution I have seen (used by DJGPP)
>>>>        
>>>>
>>>I will write that later on the wiki if you haven't before.
>>>      
>>>
>>There may be more elegant solutions, feel free to cogitate.
>>    
>>
>
>Note taken. 
>
>  
>