On 8 June 2016 at 17:14, David Moshal <dmos...@gmail.com> wrote: > David you make good points, though I have to say, sadly, that serial > vs JTag is the least of MyNewt's worries when it comes to targeting > the Maker market, if indeed that is your target market. Just saying. > > I'm happy to explain that statement further if anyone is interested. >
I'd like to hear more please, it's very relevant to aspects of my mission at the moment... To make things simpler for the Maker market I have a MicroPython prototype up and running on MyNewt. It comes with a couple of Python modules for accessing GPIO (very basic) and configuring BLE (support for: uuid16 or uuid128 bit GATT Services/Characteristics and Eddystone beacons) it also has a simple Python 'single line eval' (a not yet multiline indentation tolerant REPL) on serial. So from this perspective the only thing to flash might be the text of a Python Script in part of a flash filesystem, but, I'd still like to be able to upgrade the MicroPython 'app' using a boot loader that was always present for app recovery/reflashing. All the best Wayne > > > > On Wed, Jun 8, 2016 at 8:37 AM, David G. Simmons <santa...@mac.com> wrote: > > I’ll add a +1 here. Especially for the ‘Maker’ market, etc. where access > to a JTAG programmer may be cost prohibitive. least common denominator > (serial, typically over USB) is a great thing. > > > > Having a bootloader that will *always* run, and is accessible via > serial, makes debugging bad firmware a lot easier, makes recovering a > bricked device possible, and makes simple programming of a device easy. If > developers don’t have to re-flash a bootloader, but just send over a new > app, that makes things simple for developers. Especially for the relatively > inexperienced developer who will have a greater tendency to send over a bad > image and cause havoc on the device. I’d also advocate for a separate > process for flashing a new bootloader just to add an extra level of > protection in case someone *does* alter the bootloader code, flash it, and > it’s bad. What I’m saying is that we should make it very hard, if almost > impossible, for software to kill hardware. > > > > Best regards, > > dg > > > >> On Jun 7, 2016, at 1:37 PM, Kevin Townsend <ke...@adafruit.com> wrote: > >> > >> I think I would probably argue as well that the bootloader should be > able to run completely independent with no user image flashed, with the > ability to flash a first image over serial, but that's just my own opinion > and biased expectations in a bootloader. Obviously, I'm curious to hear > what everyone else thinks! > >> > >> There should perhaps also be an options for a fail-safe mechanism to > boot into 'bootloader only' mode (polling a pin at startup) where no > firmware is executed, but you can still talk to the bootloader via the > 'newtmgr' tool to flash an image, etc. > >> > >> BLE complicates things, but keeping 'serial' as a baseline in the core > bootloader image should add a great deal more resilience to mynewt devices, > particularly if you're devices cost several hundreds dollars per unit and > aren't just say $30 nodes that you can pull out and replace. > >> > >> K. > > > > -- > > David G. Simmons > > (919) 534-5099 > > Web • Blog • Linkedin • Twitter • GitHub > > /** Message digitally signed for security and authenticity. > > * If you cannot read the PGP.sig attachment, please go to > > * http://www.gnupg.com/ Secure your email!!! > > * Public key available at keyserver.pgp.com > > **/ > > ♺ This email uses 100% recycled electrons. Don't blow it by printing! > > > > There are only 2 hard things in computer science: Cache invalidation, > naming things, and off-by-one errors. > > > > >
