Go to the FreeLists Home Page Home Signup Help Login
 



[openbeosnetteam] || [Date Prev] [08-2004 Date Index] [Date Next] || [Thread Prev] [08-2004 Thread Index] [Thread Next]

[openbeosnetteam] Re: libnetapi.so progress

  • From: "Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx>
  • To: openbeosnetteam@xxxxxxxxxxxxx
  • Date: Sat, 14 Aug 2004 12:21:14 +0200 CEST
"Philippe Houdoin" <philippe.houdoin@xxxxxxx> wrote:
> I work a little on completing libnetapi.so.
> Alexandre Bique offered me last week to implement NetEndpoint.cpp. 
> He did it quickly, but alas without any binary compatibility in mind.
> 
> However, his contribution awake me about this task, and I try to 
> correct his implementation.
> It's now into CVS, and we now build a libnetapi.so that I still need 
> to 
> check compatibility with R5's one.
> 
> In the process, I noticed that BNetEndpoint class, and the Nettle's 
> NLEndpoint which  worked behind her in R5, assert AF_INET family 
> endpoints. 
> I think it's wrong, now that our stack support more families, 
> starting 
> with PF_ROUTE one.
> 
> That's why I've introduced a new BNetEndpoint constructor:
> BNetEndpoint(int family, int type, int protocol);
> 
> AFAIK, constructors don't count on class size, as they are not 
> virtual, 
> right?
> What do you think?
> 
> "Too early, better to fix your broken implementation" ?
> Okay, I hear you guys! ;-)

I think it would be a good addition - OTOH I've heard many complaints 
about the API to be too close to socket programming anyway, so maybe it 
would make even more sense to replace it for R2.
But that little constructor surely won't hurt :)

Bye,
   Axel.






[ Home | Signup | Help | Login | Archives | Lists ]

All trademarks and copyrights within the FreeLists archives are owned by their respective owners.
Everything else ©2007 Avenir Technologies, LLC.