Improvements to CMakeLists #2313

m00nwtchr · 2023-05-16T14:42:06Z

This PR brings (as far as i can tell) all of the functionality of the Makefile into CMake, which makes it more automated and friendly to package maintainers, and it follows the typical CMake project convention.

Example build:

$ git clone <Hyprland>
$ cmake -B build -S Hyprland
$ cmake --build build

If desired, can be installed with cmake --install build

~~It should be ready to go as is.~~
Nvm, i found one issue

m00nwtchr · 2023-05-16T16:13:46Z

So, the issue is that while this works fine on its own, for some reason under Arch's package build system I get a build error: make[2]: *** No rule to make target 'wlroots/lib/libwlroots.so.12032', needed by 'Hyprland'. Stop.

Edit: Looks like its caused by the -j option in MAKEFLAGS.

m00nwtchr · 2023-05-16T16:31:22Z

Ok, that fixed it

CMakeLists.txt

vaxerski · 2023-05-25T14:09:47Z

Would be nice if it actually succeeded through the pipelines. If you do make clear and then make release it fails because ninja can't find libwlroots, unsure why

vaxerski · 2023-05-25T18:44:14Z

Reverting my commits will not get you anywhere as this will not be merged without .pc installing to /usr/share.

m00nwtchr · 2023-05-25T18:57:25Z

The Makefile will install to /usr/share now

vaxerski · 2023-05-25T21:04:15Z

Makefile

-cleaninstall:
-	echo -en "$(MAKE) cleaninstall has been DEPRECATED, you should avoid using it in the future.\nRunning $(MAKE) install instead...\n"
-	$(MAKE) install
+	cmake --install build --prefix "/usr"


fun of you to disallow commits from me on my own project.

Hyprland installs to /usr/local, .pc to /usr.

cmake --install build if [ -d /usr/share/pkgconfig ]; then cp ./build/hyprland.pc /usr/share/pkgconfig 2>/dev/null || true; fi

if anything.

stephan-cr

@m00nwtchr I had the same idea and then I found this PR. :-) Left some comments and hope they're helpful.

stephan-cr · 2023-07-29T08:40:51Z

CMakeLists.txt

-  "protocols/")
+set(WLROOTS_SOVERSION 12032)
+
+find_program(MESON_EXE NAMES meson)


It should be checked if find_program actually finds something. Since CMake 3.18 you can say:

Suggested change

find_program(MESON_EXE NAMES meson)

find_program(MESON_EXE NAMES meson REQUIRED)

https://cmake.org/cmake/help/latest/command/find_program.html

stephan-cr · 2023-07-29T08:52:33Z

CMakeLists.txt

 )

+install(TARGETS Hyprland hyprctl DESTINATION bin)


I wonder if it would make sense to include GNUInstallDirs and doing

install(TARGETS Hyprland hyprctl DESTINATION ${CMAKE_INSTALL_BINDIR})

instead?

stephan-cr · 2023-07-29T08:59:19Z

CMakeLists.txt

+ExternalProject_Add(wlroots_build
+    PREFIX wlroots
+    SOURCE_DIR ${CMAKE_SOURCE_DIR}/subprojects/wlroots
+    PATCH_COMMAND ${SED_EXE} -E -i -e "s/(soversion = 12)([^032]|$$)/soversion = ${WLROOTS_SOVERSION}/g" ${CMAKE_SOURCE_DIR}/subprojects/wlroots/meson.build


Maybe the Sed script could be simplified to:

PATCH_COMMAND ${SED_EXE} -E -i -e "/^soversion = /s/12([^032]|$$)/${WLROOTS_SOVERSION}/" ${CMAKE_SOURCE_DIR}/subprojects/wlroots/meson.build

stephan-cr

I think I know why the build fails currently.

stephan-cr · 2023-07-29T11:11:56Z

CMakeLists.txt

+    INSTALL_COMMAND ${MESON_EXE} install --destdir ${CMAKE_CURRENT_BINARY_DIR}/wlroots
+	BUILD_BYPRODUCTS ${CMAKE_CURRENT_BINARY_DIR}/wlroots/lib/libwlroots.so.${WLROOTS_SOVERSION})
+
+set(wlroots_INCLUDE_DIR ${CMAKE_SOURCE_DIR}/subprojects/wlroots/include/)


I guess it should be something like:

ExternalProject_Get_property(wlroots_build INSTALL_DIR) set(wlroots_INCLUDE_DIR ${INSTALL_DIR}/include/)

I think that is reason why the build currently fails. config.h is a generated file and is either in the build directory or in the final "installation" directory.

stephan-cr · 2023-07-29T11:21:52Z

CMakeLists.txt

+    CONFIGURE_COMMAND ${MESON_EXE} setup --prefix "/" . ${CMAKE_SOURCE_DIR}/subprojects/wlroots
+    BUILD_COMMAND ${NINJA_EXE} -C .
+    INSTALL_COMMAND ${MESON_EXE} install --destdir ${CMAKE_CURRENT_BINARY_DIR}/wlroots
+	BUILD_BYPRODUCTS ${CMAKE_CURRENT_BINARY_DIR}/wlroots/lib/libwlroots.so.${WLROOTS_SOVERSION})


Nit: wrong indention?

majestrate · 2023-09-14T14:50:30Z

CMakeLists.txt

@@ -153,35 +172,53 @@ include(CPack)
 message(STATUS "Setting link libraries")

 function(protocol protoPath protoName external)
-    if (external)
+    if(external)
        execute_process(


using add_custom_command() here to generate these files into CMAKE_CURRENT_BINARY_DIR would be nice. this way we could add them to an interface library target that any dependents can target_link_libraries() to.

m00nwtchr · 2023-09-22T13:48:57Z

Note to anyone still looking at this: I gave up on having this upstreamed due to the above conversations. But if you still want to continue this you're welcome to fork my fork.

majestrate · 2023-09-22T14:28:47Z

Note to anyone still looking at this: I gave up on having this upstreamed due to the above conversations. But if you still want to continue this you're welcome to fork my fork.

i think the underlying pain point comes from the lack of separation in the dev branches and any packaging fixups. when packaging logic gets woven so deeply into the main development branch 100% of the time package maintainers have their fixups for their own distros mucking up things where it does not have to.
debian has spinoff branches for packages for this very reason, instead of mucking about in the dev process of everything they package they periodically rebase spinoff branches for applying small commits for JUST packaging. packaging is a full time job and making the life of the package maintainer is just as important as the people working on the bulk of the meat. when package maintainers and core software developers have to care about divergences from any distro getting in their way something is wrong with how packaging is done. if hyprland could do something more manageable like spinoff branches for nix flakes, RPMs, ebuilds, debs or whatever ever packaging is required i think that this problem would go away and everyone ends up happy as a result.
i would personally like to feel out how people in hyprland feel about that as i think that everyone who wants to contribute but doesnt is influenced by the build system being so convoluted by being intertwined with packaging right now.

m00nwtchr · 2023-09-29T07:18:16Z

Note to anyone still looking at this: I gave up on having this upstreamed due to the above conversations. But if you still want to continue this you're welcome to fork my fork.

i think the underlying pain point comes from the lack of separation in the dev branches and any packaging fixups. when packaging logic gets woven so deeply into the main development branch 100% of the time package maintainers have their fixups for their own distros mucking up things where it does not have to.
debian has spinoff branches for packages for this very reason, instead of mucking about in the dev process of everything they package they periodically rebase spinoff branches for applying small commits for JUST packaging. packaging is a full time job and making the life of the package maintainer is just as important as the people working on the bulk of the meat. when package maintainers and core software developers have to care about divergences from any distro getting in their way something is wrong with how packaging is done. if hyprland could do something more manageable like spinoff branches for nix flakes, RPMs, ebuilds, debs or whatever ever packaging is required i think that this problem would go away and everyone ends up happy as a result.
i would personally like to feel out how people in hyprland feel about that as i think that everyone who wants to contribute but doesnt is influenced by the build system being so convoluted by being intertwined with packaging right now.

That's what I tried to do, the upstream shouldn't care about the target distribution at all, and it should be left to the packagers, but they insisted on supporting distributions when installing to system with cmake, which is just moronic (and makes downstream packaging more difficult)

majestrate · 2023-09-29T14:51:03Z

Note to anyone still looking at this: I gave up on having this upstreamed due to the above conversations. But if you still want to continue this you're welcome to fork my fork.

i think the underlying pain point comes from the lack of separation in the dev branches and any packaging fixups. when packaging logic gets woven so deeply into the main development branch 100% of the time package maintainers have their fixups for their own distros mucking up things where it does not have to.
debian has spinoff branches for packages for this very reason, instead of mucking about in the dev process of everything they package they periodically rebase spinoff branches for applying small commits for JUST packaging. packaging is a full time job and making the life of the package maintainer is just as important as the people working on the bulk of the meat. when package maintainers and core software developers have to care about divergences from any distro getting in their way something is wrong with how packaging is done. if hyprland could do something more manageable like spinoff branches for nix flakes, RPMs, ebuilds, debs or whatever ever packaging is required i think that this problem would go away and everyone ends up happy as a result.
i would personally like to feel out how people in hyprland feel about that as i think that everyone who wants to contribute but doesnt is influenced by the build system being so convoluted by being intertwined with packaging right now.

That's what I tried to do, the upstream shouldn't care about the target distribution at all, and it should be left to the packagers, but they insisted on supporting distributions when installing to system with cmake, which is just moronic (and makes downstream packaging more difficult)

i personally would like to reduce the friction on this somehow so hyprland can eventually be packaged in debian.

m00nwtchr added 4 commits May 16, 2023 16:15

install rules, wlroots and udis86 auto build, hyprctl executable

10d926e

wlroots install

78080b1

change path

2d73419

Rename wlroots library target

ab30d91

m00nwtchr added 2 commits May 16, 2023 18:16

derp

f2352f9

add wlroots_build as dependency to wlroots library

00be9ab

vaxerski requested changes May 16, 2023

View reviewed changes

CMakeLists.txt Outdated Show resolved Hide resolved

CMakeLists.txt Outdated Show resolved Hide resolved

CMakeLists.txt Outdated Show resolved Hide resolved

CMakeLists.txt Show resolved Hide resolved

add plugin dev headers, only install .h

0f0a934

m00nwtchr mentioned this pull request May 17, 2023

Update PKGBUILD-git to use the fixes to CMakeLists hyprwm/aur#4

Draft

vaxerski requested changes May 22, 2023

View reviewed changes

CMakeLists.txt Show resolved Hide resolved

vaxerski added 2 commits May 25, 2023 15:53

Install .pc to /usr/share/pkgconfig

26a629b

cleanup makefile

fe7c3de

m00nwtchr added 5 commits May 25, 2023 16:46

Revert 'Install .pc to /usr/share/pkgconfig'

c706fc6

Use cmake in ci workflow

2af4ed5

Revert 'Use cmake in ci workflow'

5c4bf27

Add CMAKE_INSTALL_PREFIX to Makefile

aea2749

Fix Ninja build error

54a0955

final prefix / libwlroots not found fix

9a8534d

vaxerski requested changes May 25, 2023

View reviewed changes

vaxerski force-pushed the main branch 2 times, most recently from 09bd49b to ce9c5fd Compare July 18, 2023 22:28

stephan-cr reviewed Jul 29, 2023

View reviewed changes

stephan-cr suggested changes Jul 29, 2023

View reviewed changes

majestrate mentioned this pull request Sep 14, 2023

build system cruft #3281

Open

majestrate reviewed Sep 14, 2023

View reviewed changes

vaxerski mentioned this pull request Oct 13, 2023

CMake pipeline improvements #3564

Merged

3 tasks

vaxerski force-pushed the main branch 2 times, most recently from bb93090 to 3a189c2 Compare January 1, 2024 17:05

vaxerski force-pushed the main branch from 12db9ba to 094bce8 Compare April 5, 2024 18:43

vaxerski force-pushed the main branch 9 times, most recently from 7110a04 to 25aec3a Compare April 27, 2024 02:17

vaxerski force-pushed the main branch from 2fd5525 to 8a22692 Compare May 3, 2024 17:08

fufexan mentioned this pull request May 3, 2024

Install files using CMake, use CMake for Nix #5667

Merged

3 tasks

vaxerski force-pushed the main branch from fb471b8 to 1237732 Compare May 3, 2024 21:40

fufexan closed this in #5667 May 5, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Improvements to CMakeLists #2313

Improvements to CMakeLists #2313

m00nwtchr commented May 16, 2023 •

edited

Loading

m00nwtchr commented May 16, 2023 •

edited

Loading

m00nwtchr commented May 16, 2023

vaxerski commented May 25, 2023 •

edited

Loading

vaxerski commented May 25, 2023

m00nwtchr commented May 25, 2023

vaxerski May 25, 2023

stephan-cr left a comment

stephan-cr Jul 29, 2023

stephan-cr Jul 29, 2023 •

edited

Loading

stephan-cr Jul 29, 2023

stephan-cr left a comment

stephan-cr Jul 29, 2023

stephan-cr Jul 29, 2023

majestrate Sep 14, 2023

m00nwtchr commented Sep 22, 2023

majestrate commented Sep 22, 2023

m00nwtchr commented Sep 29, 2023 •

edited

Loading

majestrate commented Sep 29, 2023

	find_program(MESON_EXE NAMES meson)
	find_program(MESON_EXE NAMES meson REQUIRED)

Improvements to CMakeLists #2313

Improvements to CMakeLists #2313

Conversation

m00nwtchr commented May 16, 2023 • edited Loading

m00nwtchr commented May 16, 2023 • edited Loading

m00nwtchr commented May 16, 2023

vaxerski commented May 25, 2023 • edited Loading

vaxerski commented May 25, 2023

m00nwtchr commented May 25, 2023

vaxerski May 25, 2023

Choose a reason for hiding this comment

stephan-cr left a comment

Choose a reason for hiding this comment

stephan-cr Jul 29, 2023

Choose a reason for hiding this comment

stephan-cr Jul 29, 2023 • edited Loading

Choose a reason for hiding this comment

stephan-cr Jul 29, 2023

Choose a reason for hiding this comment

stephan-cr left a comment

Choose a reason for hiding this comment

stephan-cr Jul 29, 2023

Choose a reason for hiding this comment

stephan-cr Jul 29, 2023

Choose a reason for hiding this comment

majestrate Sep 14, 2023

Choose a reason for hiding this comment

m00nwtchr commented Sep 22, 2023

majestrate commented Sep 22, 2023

m00nwtchr commented Sep 29, 2023 • edited Loading

majestrate commented Sep 29, 2023

m00nwtchr commented May 16, 2023 •

edited

Loading

m00nwtchr commented May 16, 2023 •

edited

Loading

vaxerski commented May 25, 2023 •

edited

Loading

stephan-cr Jul 29, 2023 •

edited

Loading

m00nwtchr commented Sep 29, 2023 •

edited

Loading