Unified build script: dry run and unit test

[andy31415]

Problem

The execution of #8221 is opaque and not tested (except if acutally run).

Would be nice to:
a) show what manual commands could be run to build CHIP (e.g. for build environments where python is not an option or to test things manually
b) unit test that the build script can run and execute things correctly

Change overview

Builds upon #8221 to add:

• ability to list "what commands would you execute?"
• add a unit test to validate what a 'build all platform' list of commands would look like

Testing

Unit test created, also manually ran the build.

[andy31415]

Converted to draft until #8221 is merged, since otherwise the diff is too large and inconvenient to review. For reviewers: feel free to review now or wait. Sorry for the large delta.

[todo]

this is only for linux. Should be moved to 'HOST' as a platform

connectedhomeip/scripts/build/build/__init__.py

Lines 56 to 66 in f864715

	# TODO: this is only for linux. Should be moved to 'HOST' as a platform
	# to also support building on MacOS
	platforms = [Platform.LINUX]
	# at this point, at least one of 'platforms' or 'boards' is non-empty
	if not boards:
	boards = set().union(*[
	TargetRelations.BoardsForPlatform(platform) for platform in platforms
	])
	elif not platforms:
	platforms = set().union(

---

This comment was generated by todo based on a TODO comment in f864715 in #8295. cc @andy31415.

[github-actions]

Size increase report for "esp32-example-build" from e922686

File	Section	File	VM
chip-shell.elf	.flash.text	-32	-32
chip-all-clusters-app.elf	.flash.text	-16	-16

Full report output

BLOAT REPORT

Files found only in the build output:
    report.csv

Comparing ./master_artifact/chip-pigweed-app.elf and ./pull_artifact/chip-pigweed-app.elf:

sections,vmsize,filesize

Comparing ./master_artifact/chip-lock-app.elf and ./pull_artifact/chip-lock-app.elf:

sections,vmsize,filesize

Comparing ./master_artifact/chip-shell.elf and ./pull_artifact/chip-shell.elf:

sections,vmsize,filesize
[Unmapped],0,32
.flash.text,-32,-32

Comparing ./master_artifact/chip-all-clusters-app.elf and ./pull_artifact/chip-all-clusters-app.elf:

sections,vmsize,filesize
.xt.prop._ZN4chip6System5Mutex6UnlockEv,0,108
.xt.lit._ZN4chip11DeviceLayer8Internal35GenericPlatformManagerImpl_FreeRTOSINS0_19PlatformManagerImplEE14_LockChipStackEv,0,48
.xt.lit._ZN4chip11DeviceLayer8Internal35GenericPlatformManagerImpl_FreeRTOSINS0_19PlatformManagerImplEE15_StartChipTimerEj,0,48
.xt.lit._ZN4chip11DeviceLayer8Internal35GenericPlatformManagerImpl_FreeRTOSINS0_19PlatformManagerImplEE16_UnlockChipStackEv,0,48
[Unmapped],0,16
.xt.prop._ZN4chip11DeviceLayer8Internal35GenericPlatformManagerImpl_FreeRTOSINS0_19PlatformManagerImplEE14_LockChipStackEv,0,12
.xt.prop._ZN4chip11DeviceLayer8Internal35GenericPlatformManagerImpl_FreeRTOSINS0_19PlatformManagerImplEE15_StartChipTimerEj,0,12
.xt.prop._ZN4chip11DeviceLayer8Internal35GenericPlatformManagerImpl_FreeRTOSINS0_19PlatformManagerImplEE16_UnlockChipStackEv,0,12
.flash.text,-16,-16
.xt.prop._ZN4chip11DeviceLayer8Internal26GenericPlatformManagerImplINS0_19PlatformManagerImplEE14_InitChipStackEv,0,-40
.xt.lit._ZN4chip11DeviceLayer8Internal35GenericPlatformManagerImpl_FreeRTOSINS0_19PlatformManagerImplEE10_PostEventEPKNS0_15ChipDeviceEventE,0,-80
.xt.lit._ZN4chip6System5Mutex6UnlockEv,0,-128

Comparing ./master_artifact/chip-persistent-storage.elf and ./pull_artifact/chip-persistent-storage.elf:

sections,vmsize,filesize

Comparing ./master_artifact/chip-temperature-measurement-app.elf and ./pull_artifact/chip-temperature-measurement-app.elf:

sections,vmsize,filesize

[mspang]

I don't agree with writing this amount of test logic in GN actions.

Tests should be self contained, and feasible to run the tests by themselves, without needing GN to generate a complex set of command lines.

Defining the entire module as inputs to the test also seems wrong, and probably caused by not defining a package. Can we set up a proper package please?

[andy31415]

I don't agree with writing this amount of test logic in GN actions.

Tests should be self contained, and feasible to run the tests by themselves, without needing GN to generate a complex set of command lines.

Defining the entire module as inputs to the test also seems wrong, and probably caused by not defining a package. Can we set up a proper package please?

Updated to use pw_python_package. I feel a bit awkward adding a .txt as a source for that, however I could not find a way for additional deps (especially for tests) and writing the text file into the test itself seemed wrong. Let me know if this needs changes.

[andy31415]

@bzbarsky-apple @Damian-Nordic @jepenven-silabs ?