Re: [Radmind-devel] layer between libsnet and radmind

[Maarten T. <mth...@ci...>]

Hi,


> BANNER:
>     S: 200-RAP <RAP-VERSION> <HOSTNAME> <RADMIND-VERSION> radmind  
> access protocol "\r\n"
>     S: 200 COMPRESS <MECHANISM> [MECHANISM...] "\r\n"
>

These lines get read by the client in connectsn2(). So connectsn2  
needs to save these for later use. So we need a new variable to  
connectsn where it can save the compression info or other capabilities.

We can also remove the capabilities from the header and let the  
client ask for them explicitly through a new command, e.g.  
"CAPAbilities".

I favor the first option:

BANNER:
     S: 200-RAP <RAP-VERSION> <HOSTNAME> <RADMIND-VERSION> radmind  
access protocol "\r\n"
     S: 200 CAPABILITY <ITEM> [ITEM...] "\r\n"

connectsn2() calls a function extract_capabilties which puts these in  
an array and returns a pointer. Connectsn2() saves the array in an  
pointer that it receives as an argument (of course connectsn() will  
need to change too):

connectsn2( struct sockaddr_in *sin, char **capa )
)
{
   ...
   line = snet_getline_multi(...);
   capa = extract_capabilities(line);
   return ...
}

Agreed?

--
maarten