The term "benefit" is highly subjective and I wouldn't want to go into a
debate about who likes what and especially about who dislikes what and
thinks it it stupid, worthless and useless (; ...
However GCC 7 does promise some interesting things - at least
interesting for me (; This would be C++17 support, improved LTO and
better warnings. It obviously fixes some bugs from the previous releases
and I personally have a strong tendency to bump into these bugs (;
Some people may find the new "-mpure-code" option interesting - it
allows one to generate a binary which can be used with the most strict
"proprietary code protection" mechanisms found on some chips, which
cause some sections of the memory to be "execute only", which means that
all reads from that section are forbidden. This obviously also includes
reads of literals by the very same code that is placed in that region,
so such firmware needs this new special option.
There is also one more thing worth mentioning. This is the first GCC
version to support ARM microcontrollers without any patching! Previously
- for all these years - GCC had the multilib config for ARM
microcontrollers only because the people who produced binaries patched
the sources to add these multilibs. If you would take a "stock" GCC
release before version 7, then without patching all you could get is a
toolchain that produces firmware for ONE_ architecture with _ONE
configuration of thumb/ARM and ONE FPU configuration. With GCC 7 this
finally changes - if you examine the script that I posted, you'll notice
that it performs absolutely no modifications to any package it
downloads.
But - as I wrote in the first paragraph - all of that is highly
subjective. If you are perfectly happy about your current toolchain and
you have absolutely no urge to upgrade it - fine (; Then this topic is
simply not for you (; But some people really like bleeding-edge
software. I do and I think I'm not alone <:
Watch this topic
Disable multi-page view