resulting binaries (.elf files) have extra stuff about arduino_stm32 HOT 10 CLOSED

If we don't compile in the whole archive, some things stop working,

Serveral other people have tried and unfortunately failed to figure out why.

If you look back in the history for platform.txt you will see that Whole archive has been removed but had to be reinstated

I think the last reason it was reinstated was because attachTimer callbacks didn't seem to work without it

People have speculated that its something to do with weak references in the code. A lot of stuff is defined as week weak in the lower levels of the code.

Unfortunately the problem with this issue is that changing it , potentially changes everything and people get quite upset when stuff that was working, suddenly stops working.

Hence considering it only saves a few kb by removing it, the safer option is to leave it in, until someone has time to do complete testing of everything.

from arduino_stm32.

Do you have some specific code that doesn't work? Everything I tried worked. * mostly serial stuff * I'd be glad to figure it out.

from arduino_stm32.

So can we think about doing this? With the recent pin map changes you can reduce the size significantly:

Sketch uses 3,740 bytes (5%) of program storage space. Maximum is 65,536 bytes.
Global variables use 1,360 bytes of dynamic memory.

from arduino_stm32.

Rick,

Have you tested attachInterrup() your latest change, from what I recall thats the thing which broken when the whole archive flag was removed, if you look in the commit history I probably wrote the reason why I had to last re-instate the flag

from arduino_stm32.

I looked at the history on platforn.txt and didn't see any changes or comments about the change

from arduino_stm32.

88d8e2e

from arduino_stm32.

http://www.stm32duino.com/viewtopic.php?f=16&t=148#p1343 more info there about the weak references causing problems.

from arduino_stm32.

Rick,

I was hoping some other uses would take a look at the branch https://github.com/rogerclarkmelbourne/Arduino_STM32/tree/fix_weak_ref_timers and test it.

i.e it works for me, and its your fix, so it should work for you, but we are a limited subset.

However if in a few days no one has bothered.

I will just release it anyway, as I think we are both confident its no worse than we had before, and is hopefully considerably better

from arduino_stm32.

This will be actioned, its pending in a branch for testing.

I will close this issue when the branch has been merged into the master

from arduino_stm32.

fixed as part of commit
9411f03

from arduino_stm32.