mysql8: Could not run sw_vers

[josephsacco]

mysql8 does not build under MacOS 12.0.01:

databases_mysql8/mysql8/work/build
:info:configure -- CMAKE_GENERATOR: Unix Makefiles
:info:configure -- Looking for sys/types.h
:info:configure -- Looking for sys/types.h - found
:info:configure -- Looking for stdint.h
:info:configure -- Looking for stdint.h - found
:info:configure -- Looking for stddef.h
:info:configure -- Looking for stddef.h - found
:info:configure -- Check size of void *
:info:configure -- Check size of void * - done
:info:configure -- SIZEOF_VOIDP 8
:info:configure CMake Error at cmake/package_name.cmake:92 (MESSAGE):
:info:configure   Could not run sw_vers
:info:configure Call Stack (most recent call first):
:info:configure   cmake/package_name.cmake:150 (GET_PACKAGE_FILE_NAME)
:info:configure   CMakeLists.txt:754 (INCLUDE)
:info:configure -- Configuring incomplete, errors occurred!
:info:configure See also "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_databases_mysql8/mysql8/work/build/CMakeFiles/CMakeOutput.log".

Log file attached.

-Joseph

[josephsacco]

main.log

[chrstphrchvz (Christopher Chavez)]

:info:configure CMake Error at cmake/package_name.cmake:92 (MESSAGE):
:info:configure   Could not run sw_vers
:info:configure Call Stack (most recent call first):
:info:configure   cmake/package_name.cmake:150 (GET_PACKAGE_FILE_NAME)
:info:configure   CMakeLists.txt:754 (INCLUDE)
:info:configure -- Configuring incomplete, errors occurred!

I’m curious—what does the sw_vers command output on macOS Monterey, and is it somehow formatted differently from previous macOS such that it would break the ​relevant CMake code?

[josephsacco]

cavandish:~ jsacco$ sw_vers
ProductName:	macOS
ProductVersion:	12.0.1
BuildVersion:	21A559

-Joseph

[chrstphrchvz (Christopher Chavez)]

Patch to debug output from sw_vers command

[chrstphrchvz (Christopher Chavez)]

I attached a patch sw_vers_debug.diff. Joseph, could you please give it a try and report what is output?

[chrstphrchvz (Christopher Chavez)]

Can someone please assign this ticket to the maintainer (@herbygillot)?

[ryandesign (Ryan Carsten Schmidt)]

Their code is certainly very stupid. Why do they run sw_vers and then parse its output with regular expressions when they could just run sw_vers -productVersion and use its output directly? Even so, I don't see why their code does not work.

[josephsacco]

Output with patch applied:

-- CMAKE_GENERATOR: Unix Makefiles
-- Looking for sys/types.h
-- Looking for sys/types.h - found
-- Looking for stdint.h
-- Looking for stdint.h - found
-- Looking for stddef.h
-- Looking for stddef.h - found
-- Check size of void *
-- Check size of void * - done
-- SIZEOF_VOIDP 8
-- SW_VERS_OUTPUT_LIST: 'ProductName:	macOS;ProductVersion:	12.0.1;BuildVersion:	21A559'
-- SW_VERS_PRODUCTNAME: 'ProductName:	macOS'
-- SW_VERS_PRODUCTVERSION: 'ProductVersion:	12.0.1'
-- CMAKE_MATCH_1: '12'
-- CMAKE_MATCH_2: '0'
CMake Error at cmake/package_name.cmake:97 (MESSAGE):
  Could not run sw_vers
Call Stack (most recent call first):
  cmake/package_name.cmake:155 (GET_PACKAGE_FILE_NAME)
  CMakeLists.txt:754 (INCLUDE)

[chrstphrchvz (Christopher Chavez)]

Another patch to debug sw_vers command output

[chrstphrchvz (Christopher Chavez)]

Joseph, thank you for testing the patch. I believe I now know what is causing the problem. Could you please test more_sw_vers_debug.diff​? The output I expect from it is:

-- SW_VERS_OUTPUT_LIST: 'ProductName:	macOS;ProductVersion:	12.0.1;BuildVersion:	21A559'
-- SW_VERS_PRODUCTNAME: 'ProductName:	macOS'
-- SW_VERS_PRODUCTVERSION: 'ProductVersion:	12.0.1'
-- CMAKE_MATCH_1: '12'
-- CMAKE_MATCH_2: '0'
-- CMAKE_MATCH_1: TRUE
-- CMAKE_MATCH_2: FALSE

I will attach another patch with a possible fix shortly.

[josephsacco]

Confirmed...

Output with more_sw_vers_debug.diff patch applied:

-- SW_VERS_OUTPUT_LIST: 'ProductName:	macOS;ProductVersion:	12.0.1;BuildVersion:	21A559'
-- SW_VERS_PRODUCTNAME: 'ProductName:	macOS'
-- SW_VERS_PRODUCTVERSION: 'ProductVersion:	12.0.1'
-- CMAKE_MATCH_1: '12'
-- CMAKE_MATCH_2: '0'
-- CMAKE_MATCH_1: TRUE
-- CMAKE_MATCH_2: FALSE
CMake Error at cmake/package_name.cmake:107 (MESSAGE):
  Could not run sw_vers

[chrstphrchvz (Christopher Chavez)]

Patch with possible fix

[chrstphrchvz (Christopher Chavez)]

If the port configures correctly with the patch, I will open a pull request. I have not found any reports of this issue upstream, so I will likely also forward the issue and the patch upstream.

[josephsacco]

Build of mysql8 completed

cavandish:bin jsacco$ port installed | grep mysql8
  mysql8 @8.0.27_0 (active)

-Joseph

[chrstphrchvz (Christopher Chavez)]

PR opened: ​https://github.com/macports/macports-ports/pull/12810

[chrstphrchvz (Christopher Chavez)]

In d478d58f2cdcafe1d93b55340b0ff26bc3d15c78/macports-ports (master):

mysql8: fix configure on macOS 12.0

Fixes: #63726

[chrstphrchvz (Christopher Chavez)]

Reported upstream: ​https://bugs.mysql.com/bug.php?id=105464

However upstream says they cannot reproduce the issue.

[josephsacco]

Hmmm... The upstream team used cmake @3.21.3. The latest on MacPorts is cmake @3.21.4 A cmake issue perhaps?

I can do the experiment this weekend when I get some free cycles.

-Joseph

[chrstphrchvz (Christopher Chavez)]

I currently don’t have any idea why upstream doesn’t observe the issue.

Maybe give Ryan’s suggestion to instead use sw_vers -productVersion (comment:7) a try: ​https://github.com/chrstphrchvz/mysql-server/commit/65ed29a0f964.patch

[josephsacco]

Ryan's patch is, of course, a patch to follow the current patch. And yes, with both applied in sequence, mysql8 builds.

-Joseph

[josephsacco]

FWIW... Today I installed cmake @3.21.3, removed patch-package_name.cmake.diff from the Portfile and ran port configure. The configuration stage failed:

CMake Error at cmake/package_name.cmake:92 (MESSAGE):
  Could not run sw_vers
Call Stack (most recent call first):
  cmake/package_name.cmake:150 (GET_PACKAGE_FILE_NAME)
  CMakeLists.txt:754 (INCLUDE)
-- Configuring incomplete, errors occurred!

So... I have no idea what magic the upstream team performed.

-Joseph

[chrstphrchvz (Christopher Chavez)]

Apparently upstream had already fixed the issue internally using what I believe is an equivalent change: ​https://bugs.mysql.com/bug.php?id=105464#c517341

[chrstphrchvz (Christopher Chavez)]

Patch for upstream fix

[chrstphrchvz (Christopher Chavez)]

Joseph, could you please test patch-package_name.cmake.2.diff (using it instead of patch-package_name.cmake.diff)? I will then let upstream know that it works for you.

[chrstphrchvz (Christopher Chavez)]

Replying to chrstphrchvz:

Joseph, could you please test patch-package_name.cmake.2.diff (using it instead of patch-package_name.cmake.diff)? I will then let upstream know that it works for you.

No need to deactivate or uninstall mysql8. I only need to know that the configure phase works, so I would do port clean mysql8 and port -vst configure mysql8.

[josephsacco]

Christopher,

• Added patch-package_name.cmake.2.diff to files directory
• Edited Portfile to use patch-package_name.cmake.2.diff instead of patch-package_name.cmake.diff
• Ran sudo port -vst configure mysql8

--->  Applying patch-package_name.cmake.2.diff
DEBUG: Environment:
DEBUG: CC_PRINT_OPTIONS='YES'
DEBUG: CC_PRINT_OPTIONS_FILE='/opt/local/var/macports/build/_Users_jsacco_ports_databases_mysql8/mysql8/work/.CC_PRINT_OPTIONS'
DEBUG: CPATH='/opt/local/include'
DEBUG: DEVELOPER_DIR='/Library/Developer/CommandLineTools'
DEBUG: LIBRARY_PATH='/opt/local/lib'
DEBUG: MACOSX_DEPLOYMENT_TARGET='12.0'
DEBUG: SDKROOT='/Library/Developer/CommandLineTools/SDKs/MacOSX12.sdk'
Executing:  cd "/opt/local/var/macports/build/_Users_jsacco_ports_databases_mysql8/mysql8/work/mysql-8.0.27" && /usr/bin/patch -p1 < '/Users/jsacco/ports/databases/mysql8/files/patch-package_name.cmake.2.diff'
DEBUG: system:  cd "/opt/local/var/macports/build/_Users_jsacco_ports_databases_mysql8/mysql8/work/mysql-8.0.27" && /usr/bin/patch -p1 < '/Users/jsacco/ports/databases/mysql8/files/patch-package_name.cmake.2.diff'

• Configure phase completed without incident.

Onwards,

-Joseph