On Saturday, February 18, 2017 at 11:40:46 AM UTC, Schooner wrote:
>
> Hi 
>
> Just a straw poll, to see if anyone is using webtalk and if they are, 
> from packages or a RIP? 
>
> There is an opportunity to reduce the dependencies for Machinekit, by 
> making webtalk an optional extra in a build. 
> Only if selected would libwebsockets be required. 
>
> This would speed up builds slightly, plus reduce the dependencies and 
> binary footprint for the vast majority of users. 
>
> Affirmative responses only required 
>
> Thanks 
>

I would say keep it in the build, it is a possibly useful piece of 
infrastructure, I have tried using it but it did not seem very userfriendly 
and so never went further than trying to compile it.

I think there was a lot of good ideas about network connected things that 
went into machinekit early on but unfortunately there are just not enough 
people able and interested in developing them further.

The idea of machinetalk and webtalk it seemed could give something like 
ethercat ( synopsis <https://www.youtube.com/watch?v=IId15I3oFQw> ) with 
mdns discovery and protobuf allowing futureproofing of protocols. 
Unfortunately it seems tieing it all together in an easy to get started 
package has not happened. With starting new api stuff I think most 
important is showing it working with lots of easy to compile and test 
examples so developers can see it works with less than a couple of hours 
setup work. I don't know if I misunderstood something but I spent a couple 
of hours trying to get it to compile and it did not seem to do 
much(probably my own fault, but maybe others had similar problems?).

Anyways I think it should stay in the dist and someone should fix it, of 
course I have no idea how such a situation could occur as I am not going to 
do any work on fixing it, and I have found other ways to do web interfaces.

One problem that I do see is that machinetalk does not have any sort of 
realtime sync timeing or deadline handling, I don't know how such a thing 
could be implemented though it must be possible to some extent as I believe 
sercos purports to be 'realtime' though ethernet can have quite a few ms of 
transmission delay but it should be possible to calculate delays and 
workout how to degrade performance and failsafe as gracefully as possible.

Another problem I had was that all code seemed to be dependent on QT 
toolkits and python c++, I don't mind python, but most of the linuxcnc 
codebase is c which I prefer and the toolkit use glade and gtk.

I ended up implementing an inflexible text protocol that could use 
ws/twisted/zmq for transport backend and started writing a glade2html shim 
with ws4py to get multiplatform browser guis from glade and pi3d to 
implement smartphone app interfaces with gles preview.

Not got any of it to a polished working state yet but the axis browser view 
looks pretty good and works on most platforms.

Anyways I would like to use machinetalk and webtalk if someone made it 
easier to use by getting it to compile properly out of the box and it had 
some easy to use and understand example code that used only python and 
glade. Keep it in the dist if at all possible.

 

-- 
website: http://www.machinekit.io blog: http://blog.machinekit.io github: 
https://github.com/machinekit
--- 
You received this message because you are subscribed to the Google Groups 
"Machinekit" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
Visit this group at https://groups.google.com/group/machinekit.
For more options, visit https://groups.google.com/d/optout.

Reply via email to