Something has grown?

[cbbrowne]

I just had a scary moment; did a "git pull jack", "make dfu" on my favorite keymap (cbbrowne), and got the following...

Making planck/rev4 with keymap cbbrowne and target dfu

avr-gcc (GCC) 4.9.2
Copyright (C) 2014 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Size before:
text data bss dec hex filename
0 29916 0 29916 74dc planck_rev4_cbbrowne.hex

Compiling: keyboards/planck/keymaps/cbbrowne/keymap.c [OK]
Compiling: ./tmk_core/common/command.c [OK]
Linking: .build/planck_rev4_cbbrowne.elf [OK]
Creating load file for Flash: .build/planck_rev4_cbbrowne.hex [OK]

Size after:
text data bss dec hex filename
0 29916 0 29916 74dc planck_rev4_cbbrowne.hex

dfu-programmer: no device present.
Error: Bootloader not found. Trying again in 5s.
Bootloader Version: 0x00 (0)
Bootloader and code overlap.
Use --suppress-bootloader-mem to ignore
tmk_core/avr.mk:110: recipe for target 'dfu' failed

I stepped back to the last time I messed with my code, and it built/installed fine, thus:

git checkout cc7604d

HEAD is now at cc7604d... Add breathing, as initial implementation with the RAISE/LOWER keys. And throw in extra macros to shorten keymaps
root@cbbrowne:/home/cbbrowne/GitStuff/qmk_firmware/keyboards/planck/keymaps/cbbrowne# make dfu
Making planck/rev4 with keymap cbbrowne and target dfu
(note: did the actual build as cbbrowne, so all this is doing is the DFU run)

avr-gcc (GCC) 4.9.2
Copyright (C) 2014 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Size before:
text data bss dec hex filename
0 24450 0 24450 5f82 planck_rev4_cbbrowne.hex

Compiling: keyboards/planck/keymaps/cbbrowne/keymap.c [OK]
Compiling: ./tmk_core/common/command.c [OK]
Linking: .build/planck_rev4_cbbrowne.elf [OK]
Creating load file for Flash: .build/planck_rev4_cbbrowne.hex [OK]

Size after:
text data bss dec hex filename
0 24450 0 24450 5f82 planck_rev4_cbbrowne.hex

Bootloader Version: 0x00 (0)
Validating...
24450 bytes used (85.27%)

I speculate that 29K of code is too much, where 24.5K is fine.

I know I have MIDI turned off, and took recommendations (seen on Reddit threads) of adding
#define NO_DEBUG
#define NO_PRINT

Nevertheless, I speculate that some of the substantial revisions to the qmk/tmk stuff over the last couple months has made something "bloat."

Further, I suggest that maybe there should be some check within the build to make sure that the load file isn't too large.

[jackhumbert]

This is likely due to #918 - if you add API_SYSEX_ENABLE = no to your Makefile, things should drop back down.

[cbbrowne]

That works, but I'm still inclined to leave this open as an issue, as:

1. Better to have API_SYSEX_ENABLE turned off by default, turn it on only explicitly
2. It would be nice to have some size measurement to guard against oversized binaries earlier in the process

[jackhumbert]

I'm aiming for it to be enabled by default for all boards - right now it's only turned on for the Planck and the EZ. I believe the only reason it's causing trouble is because your Makefile (and a lot of others) is overriding some of the defaults that cut down on the size.

You can see the hex size when compiling (74dc, needs to be < 7000), which is an easy way to tell how big things are before trying to flash.

I know you commented on it, but this is really a dup of #920 as well.

[cbbrowne]

Yeah, it's mostly duplicative of #920, closing.