< Previous by Date Date Index Next by Date >
< Previous in Thread Thread Index Next in Thread >

Re: [reSIProcate] Request for symbols definition (WS and WSS)


On 17/09/13 17:46, Dario Bozzali wrote:
> Hi all,
> Sorry to bother you (again) about WS and WSS, but I noticed that in 
> isDgramTransport function, used in TransportSelector::getFirstInterface() 
> method (file TransportSelector.cxx), cases for WS ans WSS are missing.
> I use WIN32 so I didn't faced issues, maybe in other configurations it could 
> be a problem.
> Best regards,
> Dario.

Thanks for this feedback, I've committed a fix for that.  After we
branch v1.9, that code could move to TransportType, so I've added a
FIXME as well.

For the Symbols::WS and Symbols::WSS, I decided to include them as well,
for consistency with the other symbols entries, even if they should be
deprecated - could anybody else provide any feedback about that?

However, when looking at it again, another possible issue appeared. 
Notice that Symbols::Sip represents the URI prefix "sip"?  To follow
this pattern, we would need

Symbols::WS = "WS";    // transport, as in transport=WS
Symbols::Ws = "ws";    // URI prefix, as in ws://example.org

so it is better to encapsulate these symbols in their own types (e.g.
TransportType).  I put in the Symbols::WS and WSS, but not the prefixes
for now.

>
>
> bool isDgramTransport (TransportType type)
> {
>    static const bool unknown_transport = false;
>    switch(type)
>    {
>       case UDP:
>       case DTLS:
>       case DCCP:
>       case SCTP:
>          return   true;
>
>       case TCP:
>       case TLS:
> +++   case WS:
> +++   case WSS:
>          return   false;
>
>       default:
>          assert(unknown_transport);
>          return unknown_transport;  // !kh! just to make it compile 
> wo/warning.
>    }
> }
>
>
> -----Original Message-----
> From: resiprocate-devel-bounces@xxxxxxxxxxxxxxx 
> [mailto:resiprocate-devel-bounces@xxxxxxxxxxxxxxx] On Behalf Of Daniel Pocock
> Sent: lunedì 16 settembre 2013 20.20
> To: resiprocate-devel@xxxxxxxxxxxxxxx
> Subject: Re: [reSIProcate] Request for symbols definition (WS and WSS)
>
>
>
> On 16/09/13 18:56, Dario Bozzali wrote:
>> Hi all,
>>
>> It would be possible to add to mainline the definitions for symbols WS 
>> and WSS (see below)?
>>
>
> I remember looking at that code when doing the WebRTC work but deciding not 
> to include WS and WSS in Symbols.?xx - I think the reason I didn't include 
> them is because I couldn't find other code explicitly referring to these 
> strings and it seemed better for people to use rutil/TransportType
>
> Can anybody else comment on these definitions?  Should they be supported or 
> are they deprecated?
>
>
>>  
>>
>> Symbols.hxx:
>>
>> static const char* WS;
>>
>> static const char* WSS;
>>
>>  
>>
>> Symbols.cxx:
>>
>> const char* Symbols::WS = "WS";
>>
>> const char* Symbols::WSS = "WSS";
>>
>>  
>>
>> Moreover, in my opinion in method TransportType::isSecure() WS and WSS 
>> transports should be handled (see below).
>>
>
>
> It looks like something went wrong with those changes when the b-webrtc 
> branch was merged.
>
> 10079 fixes isSecure, I have cherry-picked it from b-webrtc into main
>
> SVN believed the change was already merged, I had to force it to cherry pick 
> this with --ignore-ancestry
>
> svn merge --ignore-ancestry -c 10079
> https://svn.resiprocate.org/rep/resiprocate/branches/b-webrtc .
>
> Thanks for the feedback about these issues
>
> Regards,
>
> Daniel
> _______________________________________________
> resiprocate-devel mailing list
> resiprocate-devel@xxxxxxxxxxxxxxx
> https://list.resiprocate.org/mailman/listinfo/resiprocate-devel
>
> _______________________________________________
> resiprocate-devel mailing list
> resiprocate-devel@xxxxxxxxxxxxxxx
> https://list.resiprocate.org/mailman/listinfo/resiprocate-devel