[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lwip-devel] Does we have a plan that merge [ppp-new] to [master]?

From: Sylvain Rochet
Subject: Re: [lwip-devel] Does we have a plan that merge [ppp-new] to [master]?
Date: Tue, 21 Aug 2012 17:52:25 +0200
User-agent: Mutt/1.5.21 (2010-09-15)


Maybe I have to give my 2 cents here too :-)

On Tue, Aug 21, 2012 at 09:19:15AM -0600, Ivan Delamer wrote:
> I'll give my 2 cents here too.
> I've been using both ppp and ppp-new. I think there is a place for both
> implementations, at least until ppp-new becomes a bit more mature.
> I suggest adding a new folder, e.g. /lwip/netif/ppp4/ or
> /lwip/netif/ppp-new/ , and perhaps a new configuration option such as
> #define LWIP_PPP_NEW or similar. You can then control which stack is being
> used through #include or makefile directives.
> Switching branches is very easy in GIT, but it takes a lot of work to keep
> the [ppp-new] branch up to date with the changes in [master]. Hence the
> suggestion.

Until now, all changes made to master and merged to the ppp-new branch 
were fast-forwarded by Git. lwIP core and PPP have great isolation 
between each other, meaning merges are easy to do.

> On another note, I have mixed feelings about the changed API. It would be
> great to keep the old function prototypes for easy porting, on the other
> hand it's good that they've changed because it forces you to revise your
> ppp code. I'm not sure if it would be possible to define some compatibility 
> macros...

That seem hard, the previous API was only allowing one PPP session, 
statically allocated at compile time in .bss. The new API is allocating 
PPP Control Block through memp.c, therefore requiring the user to pass a 
PPP control pointer to functions.

I am not saying that it is not possible, we can keep somewhere in memory 
the pointer to the unique allocatable PPP control block if the previous 
API is enabled, but the API is so simple and the change in lwIP user 
code is pretty obvious too, so I don't think this should be considered 
as a requirement.

> I think I saw some in an earlier release.

Well, I added in some place the older API, commented out with #if 0, 
this should be removed later.

> PS: ppp-new has been running for a week here with no issues.

I am using ppp-new (with PPPoE and an ADSL modem), and I did not have 
any state machine issue (reconnection after LCP timeout or after reset 
from the other side) since the great state machine rework from 
2012-07-24 and previous days commits, so it is running nicely for about 
1 month now.


Attachment: signature.asc
Description: Digital signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]