[Sur] [Server-devel] Problems with mesh OLPC Sur list / problemas con la malla

Kim Quirk kim en laptop.org
Mie Mayo 21 10:02:50 EDT 2008


Since I can't read Martin's answers in Spanish, I thought I would
answer a few in English. If these two sets of answers don't agree, can
someone point it out so we can get them right.

See inline below.

Kim


2008/5/15 Yama Ploskonka <yama en netoso.com>:
> Hi y'all, I am double posting for I do not know whose fish this is.
>
> A teacher in Uruguay indicates issues with operating with mesh, two
> others confirmed they had a similar experience of random connectivity.
>
> Please help, either subscribe to OLPC-Sur to answer in Spanish or
> respond through either list or personally and I will translate back.
>
> Yama
>  >>>>>>>>>>>>>>>>>
> Translation of the original message:
>
> The problem I was mentioning is that we know the mesh in itself to work,
> what we have not been able to do is to achieve a collaborative activity
> such as you are supposed to be able to.
> Chat works perfectly, I connect, I see who is in the neighborhood, I
> invite that person and we chat; but it is very difficult with other
> activities.
> The connection is very unstable and in a group with 15-20 children there
> is always a couple XO that will not connect or they don't see the rest
> even if they are connected.  We have solved this by working by pairs,
> but if it is supposed that the plan is one computer per child, what is
> proper is that each one have one, isn't it?
> As a sidenote, I am writing from the XO, two blocks from my house there
> is a school that has un punto de acceso, I am lucky I can connect!!!
>
>
> In the neighborhood you can see 3 points of the mesh (I do not know if
> that is what you call them, doesn't matter) 1, 6, 11.  Generally
> connection works on 1.  After this is established, it lasts less than
> half an hour and it breaks, or some XO disconnect.
> I want to make clear that even though the connection is random, I can
> choose on what mesh to connect by clicking on that point, it takes a few
> minutes but you can do it.
>
> Questions:
> Do you know how far this can be done?
==> Laptops should be able to connect to each other at 100meters or
possibly more, if there are no RF 'barriers'  (such as thick walls,
metal or screen meshes).

> Are the numbers related to that radius?
==> I think you mean the number of laptops. The more laptops that are
within the radius of the wifi, the higher the traffic and the less
reliable the connections will be. We have made a number of changes in
code and expect to make many more between now and the August release.
Also, we are recommending infrastructure access points for classrooms
with more than 20 laptops.

> Why do those XOs that are in one mesh see each other but not those in
> another?
==> The simple mesh channels 1, 6, and 11 are separate RF channels.
There is no bridging in simple mesh mode. In a school server
environment, the school server will bridge the three channels.

> It would be good that all the people who are connected see each other
> and the different points act as bridges to gather those who are farther
> away.
> In the classroom not all can connect to the same mesh so it is
> impossible to achieve a shared project.
==> If you don't have a school server, then you should have all the
children click on the same mesh channel (1, 6, or 11).

> Another matter is that, suppose we can connect several people, chat or
> talk works impeccably but when we try to share activities that are like
> more complex such as Paint or Write (don't even mention eToys) all gets
> much more difficult.

==> Yes, most of these are bugs that we are still fixing. Some fixes
were introduced in the 703 build; more bugs are being addressed in the
upcoming August release. We recognize that the reliability of mesh and
collaboration is high priority.
>
> greeTings
>
>  >>>>>>>>>>>>>>>>>>>>>
>
> original message follows > El problema al que hacía referencia es que
> conocemos el funcionamiento
>> de la malla en sí, lo que no hemos podido lograr es realizar una
>> actividad colaborativa tal como se supone que debería ser.
>> El chat funciona perfectamente. me conecto, veo quién está en el
>> vecindario, lo invito y charlamos; pero es muy difícil con las otras
>> actividades.
>> La conexión es muy inesatable y en un grupo de 15 o 20 niños siempre
>> hay un par de xo que no se conectan o no ven a las demás aunque estén
>> conectadas. Lo hemos solucionado trabajando en duplas, pero si se
>> supone que el plan es una compu por niño, lo correcto sería que cada
>> uno tenga la suya no?
>> Comento que estoy escribiendo desde la olpc, a 2 cuadras de mi casa
>> hay una escuela que tiene un access point, por suerte me puedo
>> conectar!!!
>
>> En el vecindario se ven 3 puntos de malla (no sé si así se llaman, no
>> importa) la 1, la 6 y la 11. Por lo general la conexión se logra en la
>> 1. Luego de establecida la misma, dura menos de media hora y se corta,
>> o se desconectan algunas xo.
>> Quiero aclarar que si bien la conexión es aleatoria, puedo elegir en
>> qué malla conectarme haciendo clic en el punto, toma unos minutos pero
>> se logra.
>>
>> Preguntasss:
>> Saben en qué radio de distancia funciona?
>> Los números se relacionan con este radio?
>> Por qué se ven entre sí las xo que están en la misma malla y no ven a
>> las que están en las otras?
>> Estaría bueno que se vieran todas las personas conectadas y que los
>> diferentes puntos hicieran de puente para captar a los que están más
>> lejos.
>> En el aula no todos pueden conectarse a la misma entonces es imposible
>> realizar un trabajo compartido.
>> Otro aspecto es que supongamos que logramos conectarnos una cuantas
>> personas, el chat o charla funciona impecable pero cuando intentamos
>> compartir actividades si se quiere más complejas como Pintar o
>> Escribir (ni que hablar de Etoys), todo se dificulta.
>>
>> Salu2
>
> _______________________________________________
> Server-devel mailing list
> Server-devel en lists.laptop.org
> http://lists.laptop.org/listinfo/server-devel
>


More information about the olpc-Sur mailing list