[chbot] ARM toolchain comparisons for STM32 F0, F1, and F4

[Volker Kuhlmann]

On Wed 12 Sep 2018 22:27:12 NZST +1200, Mark Atherton wrote:

> ARM do have a free toolchain gcc-arm-none-eabi (85MB) available,
> which seems odd given the conflict of interest with Keil.

The work on gcc to support ARM is probably not only from ARM, and
predates their purchase of Keil, so back then it might have been all
they had.

My experience with expensive manufacturer-supplied toolchains is that
their code size is very small (i.e. optimisation very good), at the
expense of everything else. Accepting syntactically incorrect code is
not unheard of. No need for comfortable working environments, businesses
will pay up anyway. Static analysis and warnings are non-existant. Many
companies seem to have found that they can't do without gcc one way or
another, it's a very good front-end. If your IDE of choice doesn't use
gcc it can also be useful to develop such that the same code compiles
with gcc, which can find you a lot of bugs upfront. And if you need any
kind of reproducability, like between workstations or QC, run the whole
translation process from the command line. Of course if it's just you
developing that may not be much of an issue.

> openSTM32 (owned by ac6) have an open source 'System Workbench for
> STM32' which seems bulky (!) at over 450MB.

Size does not matter... Does it work? How good is the debugger? It'll be
using gcc.

> Finally, at least the older uVision toolchain appears to allow GCC
> tools to be installed. Anyone used this combination, if there is no
> code limit size on their debugger, this would be ideal.

Some of the IDEs use open source toolchains but proprietory debuggers
that cost real $$. Do the old versions support modern parts, esp
debugging? Ask them?

Or ask the TI rep, that's what they get paid for, using their parts.

Volker

-- 
Volker Kuhlmann
http://volker.top.geek.nz/	Please do not CC list postings to me.