This patch adds support for tracking the swiftmodules for executables
exporting symbols.
This fixes a bug in the earlier implementation around emitting the
swiftmodule. Previously, the code would use
`CMAKE_EXE_EXPORTS_Swift_FLAG` to inject the `-emit-module`, and module
path information into the `CMAKE_Swift_LINK_EXECUTABLE` rule. Because
Swift skips the build step and only runs during the link phase, these
flags were injected in `cmNinjaNormalTargetGenerator::ComputeLinkCmd`
instead of `cmLocalGenerator::GetTargetFlags` where it is done normally.
Unfortunately, injecting in `ComputeLinkCmd` didn't do anything because
we have a `linkCmd` so `ComputeLinkCmd` exits early, before the
EXE_EXPORT flags get added to the link command.
Instead of playing with that flag, CMake checks
`CMAKE_Swift_LINK_EXECUTABLE_WITH_EXPORTS` and uses that as the link
rule if it exists and falls back on `CMAKE_Swift_LINK_EXECUTABLE`. I've
defined that variable in terms of `CMAKE_Swift_LINK_EXECUTABLE` with the
necessary additional flags for emitting the swift module instead. This
has the same end effect as the desired behavior, but simplifies things a
bit.
Since we're generating the swiftmodule for executables with exports,
I've also updated the dependency graph to include the swiftmodule as an
output in the build dependency graph if the executable has exports.
Tests updated:
- RunCMake/NoWorkToDo:
Ensure that the build graph does not result in unnecessary rebuilds
with exporting executables.
- SwiftOnly:
Ensure that we can consume functions defined in the executable by a
library getting linked into said executable.
Since commit 18e1bfbb3c (cmake: On configure error suggest looking at
CMake*.log files, 2013-07-09, v2.8.12~210^2), when configuring a project
fails, we print `See also .../CMake{Output,Error}.log` near the end of
the output. This was intended to help users find failures in system and
compiler inspection checks, but for normal project errors the messages
may be misleading. The logs may contain incidental errors that are part
of normal operation and do not need to be addressed by the user.
Since commit f6ed2585e5 (Modules: Record system inspection steps in the
configure log, 2023-01-16), CMake's builtin modules no longer log
information to the old-style `CMake{Output,Error}.log` files anyway,
so stop mentioning them.
Fixes: #22131
Issue: #23200
f6ed2585e5 Modules: Record system inspection steps in the configure log
0f688386ea Modules: Drop redundant check logging to CMakeOutput.log and CMakeError.log
ecc26f98eb UsewxWidgets: Remove leftover debugging code
874c2e1198 FindQt4: Drop outdated advice to look at CMakeError.log on failure
a80465bcad GHS: Drop debugging message from log
9199449687 CompileFeatures: Warn explicitly when feature detection binary is not found
24ccc8c3c9 CompilerId: Restore logging of failed identifications
95976514f6 Tests: Avoid using CMake{Output,Error}.log files
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !8089
Replace old-style `file(APPEND .../CMake{Output,Error}.log)` logging
with calls to `message(CONFIGURE_LOG)` to record the steps in the
`CMakeConfigureLog.yaml` configure log instead.
Issue: #23200
Run the `clang-format.bash` script to update all our C and C++ code to a
new style defined by `.clang-format`. Use `clang-format` version 15.
* If you reached this commit for a line in `git blame`, re-run the blame
operation starting at the parent of this commit to see older history
for the content.
* See the parent commit for instructions to rebase a change across this
style transition commit.
Fixes: #24315
a78cba5197 message: Add CONFIGURE_LOG mode to record a message in the configure log
645671d36f Help: Document configure log behavior in try_compile and try_run
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !8080
48292c8624 try_compile: Record stack of in-progess checks in configure log
d4bf7d80c6 try_compile: Add a NO_LOG option to skip recording in the configure log
9d9e8450a8 try_compile: Add optional LOG_DESCRIPTION to record in configure log
65ed5c2ca8 try_compile: Report underlying error when COPY_FILE fails
0418efb7ad Tests: Add explicit ConfigureLog case to RunCMake.try_compile
189557bd74 cmake: Make entire in-progress check stack available internally
96ce3581ab Help: Clarify backtrace order in cmake-configure-log(7)
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: buildbot <buildbot@kitware.com>
Merge-request: !8072
Many `try_compile` and `try_run` calls occur inside check modules
between `message(CHECK_START)` and `message(CHECK_{PASS,FAIL})` pairs.
Add a field to configure log entries to report this context.
Issue: #23200
The Ninja and Makefile generators honor `target_compile_options`
and friends for MASM `.asm` sources. Teach the VS generator to
honor them too for consistency.
Fixes: #24289
Update the logic added by commit ef41d49812 (Fix implicit include
directory extraction for adaptive relative paths, 2019-04-08,
v3.14.2~5^2) to account for the new `try_compile` work directory used
for the ABI check since commit 2edf0fc6d7 (Modules: Use new
keyword-dispatched try_compile signature, 2022-09-13,
v3.25.0-rc1~144^2).
Paths relative to the work directory will now have one more `../`
in them, so update the test data to match.
Fixes: #24279
Since commit aa9220d3a0 (try_compile: Add keyword-dispatched signature,
2022-09-02, v3.25.0-rc1~178^2) the `DoNotRunExecutable` code path may be
reached with no single source-file argument. Do not assume it exists.
Fixes: #24295
The default branch detection added by commit 26ec2e2b0c (Tests: Update
CTest.UpdateGIT test for custom defaultBranch, 2020-12-02,
v3.20.0-rc1~355^2) fails if the test is run inside a Git work tree whose
`.git` directory configures a `defaultBranch` that is different from the
global or system-wide value. Fix the detection to ignore the locally
configured value so that we match the `git init` decision.
5dc17d0a7a gitlab-ci: update to new Clang CI image for C++ modules
e84fcbcb0b cxxmodules: support new round of Clang patches
5e35913382 ci: update `llvm` snapshot to newest patch revisions
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: buildbot <buildbot@kitware.com>
Merge-request: !8019
In order to support multiple log versions without buffering the
entire log, move versioning to the level of individual events.
Multiple versions of an event may then be logged consecutively.
Issue: #23200
Add infrastructure for a "configure log". Use YAML for a balance of
machine- and human-readability to records details of configure-time
events in a structured format.
Teach the RunCMake test framework to support matching the configure log.
Issue: #23200
When evaluating the `$<CONFIG>` genex on an imported target has the
`IMPORTED_CONFIGURATIONS` property set, the current project has a config
that does not match any of those, and no explicit configuration mappings
are defined, fall back to the same configuration as `IMPORTED_LOCATION`.
Fixes: #24222
These patches now support the `-MF` output, so remove the `none` support
added just for the old patchset which did not use it.
Also update the flag name to `-fmodule-output=`.
Due to the new Clang module mapper flag, use a new experimental support
UUID as well.
In commit 3166547cf6 (ASM_MARMASM: Add support for Microsoft ARM
assembler language, 2022-10-14) we overlooked populating the runtime
library selection flags for the Microsoft ARM assembler, needed since
commit 0e96a20478 (MSVC: Add abstraction for debug information format,
2022-08-25, v3.25.0-rc1~142^2~1). Its compiler id is `MSVC`, so our
generators expect the table to be populated. It only supports
`Embedded` debug info. Use empty flags for the other formats.
Without this fix, enabling the `ASM_MARMASM` language with policy
`CMP0141` set to `NEW` causes an error due to the missing table entries.
Issue: #24249
Brad King