Update toolchains to C++17 #1825

mikee47 · 2019-08-29T06:36:20Z

Updates framework to build using GCC 9.2.0 toolchain for C++17.

Toolchains are at https://github.com/earlephilhower/esp-quick-toolchain/releases.

See docs/source/arch/esp8266/getting-started/eqt.rst for installation details.

(Also at https://smingdev.readthedocs.io/en/dev-esp-quick-toolchain/arch/esp8266/getting-started/eqt.html)

Extract the toolchain to a suitable location, e.g. opt/esp-quick-toolchain or C:\tools\esp-quick-toolchain, and set ESP_HOME accordingly.

The main change for the new toolchain is that the core PROGMEM handling stuff is included in sys/pgmspace.h. See esp8266/Arduino#5376 for more details.

Also includes de-duplication for NUL-terminated PSTR definitions.

…en use a link wrapper

Keeps `esp8266` Component focused on SDK Add `libc_replacements` definitions required by newlib

mikee47 · 2019-12-01T17:29:05Z

OK, ready for review.

slaff · 2019-12-02T10:17:51Z

.gitmodules

@@ -71,7 +71,7 @@

 [submodule "Host.lwip"]
 	path = Sming/Arch/Host/Components/lwip/lwip
-	url = git://git.savannah.gnu.org/lwip.git
+	url = https://git.savannah.nongnu.org/git/lwip.git


@mikee47 Why is that change here?

#1988 refers. This same repo. is used as a submodule in esp8266 LWIP, using this URL. It didn't fix the problem so can pull this commit out and deal with that separately.

so can pull this commit out and deal with that separately.

Yes, please.

.travis/build.sh

slaff · 2019-12-02T10:20:10Z

Sming/Arch/Esp8266/Components/esp8266/ESP8266_NONOS_SDK.patch

+diff --git a/lib/libc.a b/lib/libc_.a
+rename from lib/libc.a
+rename to lib/libc_.a
+diff --git a/lib/libgcc.a b/lib/libgcc_.a


Why is the underscore renaming here?

Renaming them eliminates the risk of their being mistakely picked up instead of the newlib ones.

Ok, can you rename them to lib(x).a -> lib(x).orig.a instead of underscore?

slaff · 2019-12-02T10:21:05Z

Sming/Arch/Esp8266/Components/heap/component.mk

 COMPONENT_VARS			+= UMM_POISON_CHECK
 ifeq ($(UMM_POISON_CHECK),1)
 GLOBAL_CFLAGS			+= -DUMM_POISON_CHECK
 endif

+COMPONENT_CFLAGS		+= -Wno-array-bounds


Why is that change needed?

If you comment this out then try compiling with ENABLE_CUSTOM_HEAP=1:

s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c: In function 'umm_init': s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:73:32: error: array subscript 1 is outside array bounds of 'char[1]' [-Werror=array-bounds] 73 | #define UMM_BLOCK(b) (umm_heap[b]) | ~~~~~~~~~^~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:75:24: note: in expansion of macro 'UMM_BLOCK' 75 | #define UMM_NBLOCK(b) (UMM_BLOCK(b).header.used.next) | ^~~~~~~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:231:5: note: in expansion of macro 'UMM_NBLOCK' 231 | UMM_NBLOCK(block_1th) = block_last | UMM_FREELIST_MASK; | ^~~~~~~~~~ In file included from s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:35: s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc_cfg.h:60:13: note: while referencing '_heap_start' 60 | extern char _heap_start; | ^~~~~~~~~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:73:32: error: array subscript 1 is outside array bounds of 'char[1]' [-Werror=array-bounds] 73 | #define UMM_BLOCK(b) (umm_heap[b]) | ~~~~~~~~~^~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:77:24: note: in expansion of macro 'UMM_BLOCK' 77 | #define UMM_NFREE(b) (UMM_BLOCK(b).body.free.next) | ^~~~~~~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:232:5: note: in expansion of macro 'UMM_NFREE' 232 | UMM_NFREE(block_1th) = 0; | ^~~~~~~~~ In file included from s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:35: s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc_cfg.h:60:13: note: while referencing '_heap_start' 60 | extern char _heap_start; | ^~~~~~~~~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:73:32: error: array subscript 1 is outside array bounds of 'char[1]' [-Werror=array-bounds] 73 | #define UMM_BLOCK(b) (umm_heap[b]) | ~~~~~~~~~^~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:76:24: note: in expansion of macro 'UMM_BLOCK' 76 | #define UMM_PBLOCK(b) (UMM_BLOCK(b).header.used.prev) | ^~~~~~~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:233:5: note: in expansion of macro 'UMM_PBLOCK' 233 | UMM_PBLOCK(block_1th) = block_0th; | ^~~~~~~~~~ In file included from s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:35: s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc_cfg.h:60:13: note: while referencing '_heap_start' 60 | extern char _heap_start; | ^~~~~~~~~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:73:32: error: array subscript 1 is outside array bounds of 'char[1]' [-Werror=array-bounds] 73 | #define UMM_BLOCK(b) (umm_heap[b]) | ~~~~~~~~~^~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:78:24: note: in expansion of macro 'UMM_BLOCK' 78 | #define UMM_PFREE(b) (UMM_BLOCK(b).body.free.prev) | ^~~~~~~~~ s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:234:5: note: in expansion of macro 'UMM_PFREE' 234 | UMM_PFREE(block_1th) = block_0th; | ^~~~~~~~~ In file included from s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc.c:35: s:/sandboxes/sming-dev/Sming/Arch/Esp8266/Components/heap/umm_malloc/src/umm_malloc_cfg.h:60:13: note: while referencing '_heap_start' 60 | extern char _heap_start; | ^~~~~~~~~~~ cc1.exe: all warnings being treated as errors

slaff · 2019-12-02T11:01:11Z

OK, ready for review.

Looks good to me. Newer language features for a small price. Here is the comparison that I've done for one of my applications (same laptop, same Sming version and same SDK version):

Roughly 300 bytes less free RAM, 50 bytes less IRAM, and 4K bigger file size.

Details below:

OLD toolchain

Memory Usage

   Section|                   Description| Start (hex)|   End (hex)|Used space
------------------------------------------------------------------------------
      data|        Initialized Data (RAM)|    3FFE8000|    3FFE8580|    1408
    rodata|           ReadOnly Data (RAM)|    3FFE8580|    3FFE9880|    4864
       bss|      Uninitialized Data (RAM)|    3FFE9880|    3FFF0B18|   29336
      text|            Cached Code (IRAM)|    40100000|    40106C50|   27728
irom0_text|           Uncached Code (SPI)|    40202010|    4027C9FD|  502253
Total Used RAM : 35608
Free RAM : 46312
Free IRam : 5058
ESPTOOL2 out/Esp8266/debug/firmware/rom0.bin

File size

524K for out/Esp8266/debug/firmware/rom0.bin

New toolchain

Memory Usage

   Section|                   Description| Start (hex)|   End (hex)|Used space
------------------------------------------------------------------------------
      data|        Initialized Data (RAM)|    3FFE8000|    3FFE8680|    1664
    rodata|           ReadOnly Data (RAM)|    3FFE8680|    3FFE99C8|    4936
       bss|      Uninitialized Data (RAM)|    3FFE99C8|    3FFF0C40|   29304
      text|            Cached Code (IRAM)|    40100000|    40106C88|   27784
irom0_text|           Uncached Code (SPI)|    40202010|    4027D5FC|  505324
Total Used RAM : 35904
Free RAM : 46016
Free IRam : 5002
ESPTOOL2 out/Esp8266/debug/firmware/rom0.bin

File size

528K for out/Esp8266/debug/firmware/rom0.bin

mikee47 · 2019-12-02T11:09:19Z

Newer language features for a small price.

Plus the compilers are now consistent for all supported platforms (I'm assuming MacOS uses the Linux toolchain?). I'll be especially happy to be rid of GCC 4.8.5 as it's problematic.

Sming/Arch/Esp8266/Components/esp8266/esp_cplusplus.cpp

Sming/Arch/Esp8266/Components/gdbstub/gdbstub-internal.h

Sming/Arch/Esp8266/Compiler/ld/common.ld

Move library binaries into `libc` Component Fix linker map Rename conflicting libraries in SDK - not used

mikee47 · 2019-12-02T11:53:15Z

Sming/build.mk

+else
+CPP_STD				:= c++17
+endif
+CXXFLAGS			+= -std=$(CPP_STD)


Note that I've elected to select C++17 for the current toolchains (except linux which doesn't support it). This has the potential to break some existing code, but AFAIK only insofar as it includes better checks and so not a bad thing IMHO. The alternative is to select C++17 for the new toolchain only. This could also be made user-selectable, but not sure that's necessarily a good move.

.. although being able to select the standard would be useful for testing against C++20, etc.

I've added support for this and included note in documentation.

mikee47 · 2019-12-02T11:57:08Z

Another thing to consider is where the toolchains live. At the moment they're pre-release, but perhaps before releasing Sming 4.0.1 we might consider adding them to our SmingTools repo. and pulling them from there.

mikee47 · 2019-12-02T11:57:59Z

I'll also add some documentation updates covering this PR.

mikee47 · 2019-12-02T12:02:13Z

Sming/Arch/Esp8266/Components/libc/component.mk

+COMPONENT_SRCFILES		+= pgmspace.c
+LIBDIRS					+= $(COMPONENT_PATH)/lib
+EXTRA_LIBS				+= microc microgcc setjmp
+endif


You may have spotted that stdc++ has been removed from the list of libraries. One reason is that it contains allocation operators (new, delete, etc.) which conflict with our own. We don't use STL streams or a bunch of other stuff that it might be required for so leaving it out simplifies things. We can always revisit in the future if it's required for any other reason.

Hello! I think that stdc++ is worth to keep. If there is any chance to do so. I use std::vector in my code and plan to switch to std:array all plain array. STL itself is good thing if used properly. Also there are libs that depend on STL flatbuffers as example. So do please keep this in Sming if possible.

The STL shouldn't need the library as it's all templated code. It was only added in #1734 because of an issue with placeholders when compiling without optimisations.

It's probable that at least some parts of libstdc++.a will be required to support exception handling, RTTI, iostream, etc. but at present Sming doesn't use those.

Would you try some test compiles with your code to see if it builds OK without the library? Just remove the stdc++ entry from Arch/Esp8266/app.mk. Post back if it turns out there is stuff you do need from there then we can address that.

Note: Host code builds in GDB mode at -O0 but as we've since updated Windows from GCC 5 to 8.2 it is no longer an issue.

slaff · 2019-12-09T13:11:55Z

@mikee47 In addition it seems that you have done great job improving also the memory usage. It looks like the new toolchain not only provides new C++ standards but also decreases the ROM size and the RAM and IRAM usage giving us additional room to expand our Sming applications.

	Old	New	Free
RAM	46604	46964	+360	more is better
IRAM	5057	6230	+1173 !!!	more is better
Rom File Size	524K	485K	- 1009	less is better

NEW

mcu2: Linking out/Esp8266/debug/build/app_0.out
   Section|                   Description| Start (hex)|   End (hex)|Used space
------------------------------------------------------------------------------
      data|        Initialized Data (RAM)|    3FFE8000|    3FFE8680|    1664
    rodata|           ReadOnly Data (RAM)|    3FFE8680|    3FFE961C|    3996
       bss|      Uninitialized Data (RAM)|    3FFE9620|    3FFF0890|   29296
      text|            Cached Code (IRAM)|    40100000|    401067BC|   26556
irom0_text|           Uncached Code (SPI)|    40202010|    402734E8|  464088
Total Used RAM : 34956
Free RAM : 46964
Free IRam : 6230

OLD

mcu2: Linking out/Esp8266/debug/build/app_0.out
   Section|                   Description| Start (hex)|   End (hex)|Used space
------------------------------------------------------------------------------
      data|        Initialized Data (RAM)|    3FFE8000|    3FFE8580|    1408
    rodata|           ReadOnly Data (RAM)|    3FFE8580|    3FFE975C|    4572
       bss|      Uninitialized Data (RAM)|    3FFE9760|    3FFF09F8|   29336
      text|            Cached Code (IRAM)|    40100000|    40106C51|   27729
irom0_text|           Uncached Code (SPI)|    40202010|    4027CA10|  502272
Total Used RAM : 35316
Free RAM : 46604
Free IRam : 5057
ESPTOOL2 out/Esp8266/debug/firmware/rom0.bin

Remove ICACHE_FLASH_ATTR definition, conflicts with one in esp_attr.h. **Don't disable interrupts during heap operations.** Sming is single-threaded and heap allocations are prohibited from interrupt context, therefore disabling interrupts during heap allocation operations is not necessary and detrimental to performance.

…rogcc.a as well (old toolchain)

slaff · 2019-12-10T10:32:34Z

Custom heap and axTLS don't play nice together with this PR. The problem is not reproducible in develop with the old 4.8 gcc toolchain.

@mikee47 Would you have time to look at this issue or you would prefer to have this PR merged and address the issue in a separate PR?

mikee47 · 2019-12-10T10:42:15Z

calloc is one of those functions which isn't used a great deal - only within axtls it seems!

When I built Basic_Ssl as you indicated I got a conflict error with ICACHE_FLASH_ATTR (now fixed) - hopefully you got that as well otherwise there's something really weird going on!

slaff · 2019-12-10T10:55:49Z

hopefully you got that

@mikee47 Thanks a lot! Basic_Ssl with Umm Malloc compiles now as expected.

mikee47 · 2019-12-10T11:01:43Z

Have you tested out GDB with the new toolchain? It includes the un-patched version which sends more stuff over the wire, but should not make any difference to how its used.

slaff · 2019-12-10T11:03:15Z

.travis.yml

+    - SDK_VERSION=3.0.1
+    - ESP_HOME=$TRAVIS_BUILD_DIR/opt/esp-alt-sdk
+
+  - stage: build c++17


@mikee47 Can you try to use build as a stage name also here? It would be better to have the different builds running in parallel and not wait for build to finish in order for build c++17 to start. I am not sure if travis will merge them under one stage and run them in parallel but it is worth trying.

mikee47 · 2019-12-10T11:38:21Z

I've attempted to test the custom heap on hardware, but it appears to be broken... Investigating....

mikee47 · 2019-12-10T13:06:32Z

I've reverted to #1745 and still cannot get even Basic_Blink to work using custom heap. Very odd.

slaff · 2019-12-10T13:23:09Z

still cannot get even Basic_Blink to work using custom heap. Very odd.

make dist-clean
cd ../samples/Basic_Blink
make config-clean # !!!
make ENABLE_CUSTOM_HEAP=1

I will try it too...

slaff · 2019-12-10T13:42:38Z

I will try it too...

Hm... I have compiled Basic_Blink and Basic_Serial with ENABLE_CUSTOM_HEAP=1 and both are just not starting for me. I can see that rBoot is up and running but the application itself does not get started at all. And to make things worse there is no error message.

mikee47 · 2019-12-10T13:56:57Z

Same here. I have two thoughts:

UMM Malloc isn't getting initialised, perhaps some static data isn't zeroed out at startup (i.e. called before BSS is initialised)
Some code may be in the wrong section. I've whacked IRAM_ATTR on all the umm malloc code, no difference.

mikee47 · 2019-12-10T13:58:34Z

What's really bugging me is that it the original commit (1745) which was supposed to fix the custom heap doesn't work any more, which might support (1) above.

slaff · 2019-12-10T14:00:34Z

Ok, add the parallel build change. I will merge this PR and then you can open a new PR related to umm_malloc.

New features ------------------ - No-WiFi build option SmingHub#2004 - get more resources if your application is not using WIFI. - Multiple SSL adapters based on axTLS and BearSSL. SmingHub#1999 - Added basic Crypto support library SmingHub#2014 - Updates framework to build using GCC 9.2.0 toolchain for C++17. SmingHub#1825 - Modbus master SmingHub#1992 - Implemented Small String Optimisation (SSO). SmingHub#1951 - Webcam stream and sample webcam web server. SmingHub#1981 - Allow HTTP connections to ignore rejected body content SmingHub#1928 Improvements ------------------- - Some improvements to multipart parser SmingHub#2007 - Update ArduinoJson to 6.13.0 SmingHub#1979 - Added precaching from Arduino for ESP8266. SmingHub#1965 - Add support for 'Expect: 100-continue' in HTTP server. SmingHub#1931 - Upgrade to FlashString Library SmingHub#1974, SmingHub#2013 Bug fixes ------------- - Updated mqtt-codec to allow publish messages without payload. SmingHub#1976 - HttpConnection freed twice. SmingHub#1938 - Hangs at startup when custom heap enabled. SmingHub#1996 - Fix issues reported by valgrind SmingHub#2017 Breaking changes and Migration ------------------------------------------- - See our [dedicated page](https://sming.readthedocs.io/en/latest/upgrading/4.0-4.1.html) for migration from 4.0.0 to 4.1.0. All PRs scheduled for this release can be seen from [here](https://github.com/SmingHub/Sming/milestone/23)

mikee47 mentioned this pull request Aug 29, 2019

Discussion: Adding C++14/17 support with -std=c++1z #1134

Closed

mikee47 force-pushed the dev/esp-quick-toolchain branch from cee146a to 83a0e4c Compare November 29, 2019 22:37

mikee47 changed the title ~~[Experimental, WIP] Build using esp-quick-toolchain~~ [WIP] Build using esp-quick-toolchain Nov 29, 2019

mikee47 force-pushed the dev/esp-quick-toolchain branch 3 times, most recently from 54f318d to 23cec4a Compare November 30, 2019 15:02

mikee47 mentioned this pull request Nov 30, 2019

LWIP2 clone failing #1988

Open

mikee47 force-pushed the dev/esp-quick-toolchain branch from 23cec4a to bfface8 Compare November 30, 2019 15:37

mikee47 changed the title ~~[WIP] Build using esp-quick-toolchain~~ [WIP] Update toolchains to C++17 Nov 30, 2019

mikee47 added 5 commits December 1, 2019 16:35

Fix strncpy issue in esptool2

5f65074

Fix malloc_count function signatures

1f28c3f

user_init() shouldn't be weak; if applications wish to override it th…

62a4ed2

…en use a link wrapper

Add GCC_VERSION build variable

a6a43d9

Move C/C++ library-related code into libc Component

5f4e70e

Keeps `esp8266` Component focused on SDK Add `libc_replacements` definitions required by newlib

mikee47 force-pushed the dev/esp-quick-toolchain branch from bfface8 to 997f2e4 Compare December 1, 2019 16:37

mikee47 changed the title ~~[WIP] Update toolchains to C++17~~ Update toolchains to C++17 Dec 1, 2019

slaff added this to the 4.0.1 milestone Dec 2, 2019

slaff added the 3 - Review label Dec 2, 2019

slaff reviewed Dec 2, 2019

View reviewed changes

mikee47 commented Dec 2, 2019

View reviewed changes

Sming/Arch/Esp8266/Components/esp8266/esp_cplusplus.cpp Outdated Show resolved Hide resolved

mikee47 commented Dec 2, 2019

View reviewed changes

Sming/Arch/Esp8266/Components/gdbstub/gdbstub-internal.h Show resolved Hide resolved

mikee47 commented Dec 2, 2019

View reviewed changes

Sming/Arch/Esp8266/Compiler/ld/common.ld Show resolved Hide resolved

Add newlib libraries for use with new toolchain

de100cb

Move library binaries into `libc` Component Fix linker map Rename conflicting libraries in SDK - not used

mikee47 commented Dec 2, 2019

View reviewed changes

mikee47 added 3 commits December 10, 2019 09:56

Fix missing calloc when building with ENABLE_CUSTOM_HEAP=1

664c745

Widen irom0 section filter to move floating point routines for libmic…

d781c6a

…rogcc.a as well (old toolchain)

slaff removed the 3 - Review label Dec 10, 2019

slaff reviewed Dec 10, 2019

View reviewed changes

Run Travis CI build stages in parallel

ad6ae3a

mikee47 mentioned this pull request Dec 10, 2019

Custom heap broken #1994

Closed

slaff merged commit 3a99771 into SmingHub:develop Dec 10, 2019

slaff added 3 - Review and removed 0 - Backlog 3 - Review labels Dec 10, 2019

mikee47 deleted the dev/esp-quick-toolchain branch December 10, 2019 14:50

mikee47 mentioned this pull request Dec 13, 2019

How to use std::exception? #1973

Closed

slaff mentioned this pull request Jan 2, 2020

Preparation for Release 4.1.0 #1941

Closed

4 tasks

mikee47 restored the dev/esp-quick-toolchain branch November 11, 2020 06:31

mikee47 deleted the dev/esp-quick-toolchain branch March 29, 2021 16:56

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update toolchains to C++17 #1825

Update toolchains to C++17 #1825

mikee47 commented Aug 29, 2019 •

edited

mikee47 commented Dec 1, 2019

slaff Dec 2, 2019

mikee47 Dec 2, 2019

slaff Dec 2, 2019

slaff Dec 2, 2019

mikee47 Dec 2, 2019

slaff Dec 2, 2019

slaff Dec 2, 2019

mikee47 Dec 2, 2019

slaff commented Dec 2, 2019

mikee47 commented Dec 2, 2019

mikee47 Dec 2, 2019

mikee47 Dec 2, 2019

mikee47 Dec 2, 2019

mikee47 commented Dec 2, 2019

mikee47 commented Dec 2, 2019

mikee47 Dec 2, 2019

avr39-ripe Dec 4, 2019

mikee47 Dec 4, 2019 •

edited

mikee47 Dec 4, 2019 •

edited

slaff commented Dec 9, 2019 •

edited

slaff commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff Dec 10, 2019 •

edited

mikee47 commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff commented Dec 10, 2019

slaff commented Dec 10, 2019 •

edited

mikee47 commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff commented Dec 10, 2019

Update toolchains to C++17 #1825

Update toolchains to C++17 #1825

Conversation

mikee47 commented Aug 29, 2019 • edited

mikee47 commented Dec 1, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

slaff commented Dec 2, 2019

OLD toolchain

Memory Usage

File size

New toolchain

Memory Usage

File size

mikee47 commented Dec 2, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mikee47 commented Dec 2, 2019

mikee47 commented Dec 2, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mikee47 Dec 4, 2019 • edited

Choose a reason for hiding this comment

mikee47 Dec 4, 2019 • edited

Choose a reason for hiding this comment

slaff commented Dec 9, 2019 • edited

NEW

OLD

slaff commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff Dec 10, 2019 • edited

Choose a reason for hiding this comment

mikee47 commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff commented Dec 10, 2019

slaff commented Dec 10, 2019 • edited

mikee47 commented Dec 10, 2019

mikee47 commented Dec 10, 2019

slaff commented Dec 10, 2019

mikee47 commented Aug 29, 2019 •

edited

mikee47 Dec 4, 2019 •

edited

mikee47 Dec 4, 2019 •

edited

slaff commented Dec 9, 2019 •

edited

slaff Dec 10, 2019 •

edited

slaff commented Dec 10, 2019 •

edited