<brief promo plug>
Especially in the latter case of developing nations - but also for mobile
users in general - enabling offline modes with synchronisation after an
indefinite period is a must.
</brief promo plug>
John

All the best,

John Blossom

email: jblos...@gmail.com
phone: 203.293.8511
google+: https://google.com/+JohnBlossom


On Tue, Jul 16, 2013 at 9:12 AM, Vicente J. Ruiz Jurado <v...@ourproject.org
> wrote:

> El 16/07/13 14:04, Ali Lown escribió:
> > On 16 July 2013 12:48, Vicente J. Ruiz Jurado <v...@ourproject.org>
> wrote:
> >> El 15/07/13 19:12, Ali Lown escribió:
> >>> Deletion:
> >>> I feel that none of these libraries serve any purpose now, and should
> >>> be removed.
> >>> - codegen/SocketIO (superseded by Websockets, could-be-argued to keep
> >>> for legacy situations, but I don't feel it is worth the maintenance
> >>> effort)
> >>> - runtime/SocketIO ditto
> >>
> >> When websocket it not available or is not working, the webclient uses
> >> socketio as a fallback (many times, because the client is under a proxy
> >> that blocks websocket connections, or because the browser is not
> >> up-to-date).
> >
> > I am aware of the uses of Socket.IO, but fail to see them as worth
> supporting:
> > 1) Outdated browser -> Why maintain support for out-of-date browsers?
> > This is the IE6 argument again...
> > 2) 'Proxy' -> If by this you mean an open 'web proxy' - why are we
> > supporting that?
> >                   Anybody attempting to 'hide' their IP would be using
> > a VPN which does not have this problem...
> >                   Are there any other uses for a 'web proxy'?
> >               -> If by this you mean an out-of-date Corporate proxy -
> > why are we supporting this case?
> >                   (IIS8 added support for Websockets for all modern
> > corporations stuck using Windows, Squid 2.X+3.X support Websockets
> > [except in 'interception mode'])
> >                   [Use of SSL will get Websockets through all but the
> > meanest of firewalls, since the only possible situation breaking the
> > packets then is a firewall performing DPI which is manipulating SSL'd
> > packets on-the-fly!]
> >                   If you are using Wave from behind your corporate
> > firewall there are 2 situations:
> >                   1) It is company-related work -> In which case they
> > will have already done the work to get Wave through their firewall
> >                   2) It is NOT company-related work -> In which case
> > why is it being done at their workplace, and why should we support it?
> >
> > Ali
> >
> > PS. Thanks for the patch, depending on the response to this, I may
> > need to use it.
> >
>
> With some of our public wave servers (for instance kune.cc) we had
> received feedback of users with connections issues:
> - from corporations using hardware and/or software not websocket capable
> (I think that even Yuri was suffering this problem in his job)
> - from universities and libraries with hard restrictions on their users
> Internet connections
> - from users in poor countries (in Central America and Asia) where
> Internet providers use proxies to minimize the bandwidth usage.
>
> Also some browsers of tablets & mobiles don't have correct websocket
> support.
>
> This is why I implemented the fallback use of socketio and I try to  fix
> the bugs I found in socketio, because we see that users from public wave
> servers use/need it. So from our point of view and from our daily
> experience a websocket fallback is very needed.
>
> I was looking for an alternative to socketio several times (atmospehere,
> cometd, gnisio, etc), but I couldn’t provide yet a patch with a more
> updated alternative to socketio-gwt.
>
> BR,
> --
> Vicente
>

Reply via email to