On Monday, 8 October 2018 12:10:25 CEST guifipedro wrote:
At the moment what I
think is: this is too much complexity. I know openwrt works the same
way with code written in C. Templates architectures, devices, etc.
Libremesh does the same but with lua code? I have enough with openwrt
complexity. And most importantly, I want users interact sharing their
config files. So I prefer a blind approach: "this options works,
include it, or I include myself".
You are saying that is not justifiable because libremesh does the
same. I did an effort to understand libremesh, did you try to
understand temba? I don't have so much documentation but essentially:
erb templates on config files per device (some files are shared)
evaluated from inherited yaml files with two interfaces: rake and
rubyonrails form.
Are you joking?
You are arguing that libremesh is too complex because it is written in Lua
(one of the most minimal and simple languages out there) and then you propose
to solve this complexity by providing yet another one man firmware with no
documentation + erb templates + yaml + rake + ruby on rails ?
I think this is too much even for Kolmogorov...
Also from what you written you seems to describe openwrt options like as if
they where just bricks of different colors which has almost no interactions
between each others, so one need just pick another if per doesn't like the
color of this, and thinks will works out of the box just with a different sum
of colors...
While I am quite confident to say it is quite common experience that the
options have very complex interactions so most of them need to be set in a
coherent way according with the others and tested with much care, and this one
of the big source of complexity not just of libremesh, but of any other
solution to similar problem.
Do you understand now how different are the firmwares
and that they
are trying to solve the problem with very different approach?
I think that Temba and libremesh are completely different beasts.
I have no problem with you trying different approaches and actually I think it
is good for you to write your own tools, but I think the problem you are
trying to solve is obviously different from the one libremesh targets, and it
is out of scope on lime-* mailing list, especially on an hijacked thread.
Cheers!
Gio