Attempt to solve Issue #26 (Motes change names while programming)

[comawill]

This PR solves #26 in different ways

1. introducing a udev-rule which generates systemlinks for nodes at /dev/inga/node-* and uses them instead of ttyUSB* (no more automatic renaming)
2. introducing a way to get static handles for the ttyUSB* by using /dev/serial/by-id/* for setups without this udev-rule
3. fallback mode to ttyUSB* if all of them fail

[ibr-jenkins]

Can one of the admins verify this patch?

[yschroeder]

you have to glob here on "/dev/ttyUSB*"

[yschroeder]

I will test this tomorrow.

[comawill]

@yschroeder thank you (fixed that typo)

[yschroeder]

We should think about how specific the udev rule should be:

In the current version we map all FTDI devices to the INGA folder. This comes in handy for motes where the initial setup wasn't done (yet). But maybe it is better to only map devices into that folder that have

ATTRS{product}=="INGA"

set.

This can prevent confusion if multiple different FTDI devices are connected. Every INGA is an FTDI but not every FTDI is an INGA 😄

[yschroeder]

I just tested this (without parallel execution as PR #27 was missing).

For me it works by getting the nodes from the /dev/inga folder. I have not tested the fallback paths.

Maybe we should glob onto /dev/inga/* instead of /dev/inga/node-*, this way it would also work for @ejoerns udev rule where the nodes have different names.

But after all: 👍

[comawill]

Changed those things:

1. only symlinks for INGA devices
2. no mode modifications
3. changed first glob to /dev/inga/* (if there is a need to be more specific, this could be changed in the future, too)

[ejoerns]

👍 looks great!

I'm not sure if platform is the right place for the udev rules or if we should place it in examples or tools. But a more important further step would be to add a hint to use it somewhere in the documentation; either in the Wiki, in a README.md or the Doxygen page, or simply everywhere ;) But this is a more general issue and does not have to be covered by this PR.