Re: [lwip-devel] Closing connections with rawapi

[Per Knytt]

On Wed, Apr 30, 2014 at 7:23 PM, Sergio R. Caprile <address@hidden> wrote:

(Based on the no response status)
I think you should ask these questions on the users list, you'll have
more attention (and luck) there.

Thanks, but I'm perhaps being too polite :)

I would say that this is definitely a bug but the question is if it's a bug in the documentation

or in the code.

I see three alternatives

1. It is forbidden to call tcp_abort() on a pcb that tcp_connect() has failed on.

    If this is true, rawapi.txt should say so.

    This would make the API more cumbersome to use since if tcp_connect() fails,

    for example because there's no free memory, then tcp_close() can fail for the same

    reason. If we can't call tcp_abort() in that situation the application needs to save state

    to be able to retry tcp_close() later until it succeeds. This is cumbersome in an

    environment that doesn't support blocking operations.

2. It is forbidden to call tcp_connect() on a pcb on which tcp_bind() has been called.

    If this is true, rawapi.txt should say so.

    I guess that this is not entirely unreasonable but a little surprising as the operations

    seems somewhat orthogonal.

3. tcp_abort() should check if the pcb is bound and in that case remove it from tcp_bound_pcbs.

    I would prefer this because it is the least surprising behaviour.

I can submit a patch for either of these three options but I'd like some indication of which is

the preferred solution. Alternatively do nothing if I'm on drugs and this is a non-issue...

Sincerely

Per Knytt