first pass at saving history out to file on exit

[jacobrosenthal]

feel free to bike shed

[Yatekii]

Thanks for providing this implementation.
I think this feature is direly needed!

What I would change tho is this: Instead of just enabling logs and have cargo.embed clutter your entire source dir with logs, we could make it such that one can provide an RTT log directory so the config flag would be of type Option<Path>.
And then we could structure the logs on a per run basis. So could make a structure like this:

rtt_log_dir
    binary1_target1_timestamp1
    binary1_target1_timestamp2
    binary1_target2_timestamp1
    binary2_target1_timestamp1

etc.

What do you think? :)

[jacobrosenthal]

defaults path to ./logs
Im generating the timestamp in main and sending it into app::new so we get a single timestamp

So with my two channels Im gettng
logs
magic_wand_2020-07-20T15:18:25.458863350-07:00_0.txt
magic_wand_2020-07-20T15:18:25.458863350-07:00_1.txt

The rtt channel is kinda getting lost in the back of the date. Also I cant seem to find the target.. And im stealing the name from the args, not sure thats the best place, presumably its being parsed somewhere else?

[jacobrosenthal]

Made channel more obvious
Could probably cut time down more but Ive got no opinion to what.
Not sure if you meant chipname when you said target but its available so..

magic_wand_STM32F407VGTx_2020-07-25T18:12:27.848447655-07:00_channel0.txt
magic_wand_STM32F407VGTx_2020-07-25T18:12:27.848447655-07:00_channel1.txt

[jacobrosenthal]

Feedback requested

[Yatekii]

LGTM now :) Thanks for doing this and sorry it took so long!

[Yatekii]

bors r+

[bors]

Build succeeded:

• Check (ubuntu-latest)
• Check (windows-latest)
• Rustfmt
• Test Suite