Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

test cases/linuxlike/14 static dynamic linkage: Add matrix by static #13842

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

test cases/linuxlike/14 static dynamic linkage: Add matrix by static #13842

wants to merge 1 commit into from
+88 −66

Conversation

dememax
Copy link

@dememax dememax commented Oct 30, 2024 •
edited
Loading

Not only Solaris doesn't ship static libraries, it's common practice for Linux distributions in general (security updates, size/space, etc.).
See, for example 'test cases/frameworks/1 boost'.

Enhancements:

  • Added flexibility: Introduced the SKIP_STATIC_ZLIB environment variable to optionally skip static linkage (similar to SKIP_STATIC_BOOST).
  • Expanded testing: Included testing for dynamic linkage, which was previously limited to static in verify_static.py (renamed to verify_zlib_linkage.py).
  • Improved test coverage: Ensured that each case is tested for its specific linkage type and the opposite case fails.
  • Addressed Cygwin issue: Ensured that the corresponding symbol is present on Cygwin.

Renames, comments, etc.

@dememax
Copy link
Author

dememax commented Oct 30, 2024 •
edited
Loading

https://github.com/mesonbuild/meson/actions/runs/11594506125/job/32280901101#step:5:587

Total failed tests:  1

All failures:
  -> common: 273 both libraries    (default_library=both default_both_libraries=static use_dep=True)

Is it my fault?

@dememax
Copy link
Author

dememax commented Oct 30, 2024

During "cygwin / gccx64ninja (pull_request) Failing", with the help of debug print, I've found:

For dynamic linkage (static=false), nm output in "linuxlike / static dynamic linkage" filtered by "zlibVersion" is:

0000000100408150 I __imp_zlibVersion
00000001004010d0 T zlibVersion

For static, it doesn't contain the symbol with "_imp".

@eli-schwartz
Copy link
Member

Small nit: commit messages should use imperative mood: https://cbea.ms/git-commit/#imperative

@dememax
Copy link
Author

dememax commented Nov 2, 2024

@eli-schwartz Done! Thanks!

@dememax dememax changed the title test cases/linuxlike/14 static dynamic linkage: Adding matrix by static test cases/linuxlike/14 static dynamic linkage: Add matrix by static Nov 2, 2024
eli-schwartz
eli-schwartz reviewed Nov 5, 2024
View reviewed changes
test cases/linuxlike/14 static dynamic linkage/meson.build Outdated
else
has_static = true
endif
project('zlib linkage', 'c')
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The project description doesn't really need to change here (it is still testing static dynamic linkage, not zlib specifically, anyway :))

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Shall be fixed! Thanks.

test cases/linuxlike/14 static dynamic linkage/meson.build
Comment on lines -3 to -4
# Solaris does not ship static libraries
if host_machine.system() == 'sunos'
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should probably still explicitly skip on Solaris. Your new approach would cause test failure if SKIP_STATIC_ZLIB is not exported, even though we know that if s and host_machine.system() == 'sunos' there will not be a static library.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You are perfectly right that this changes the behavior, and it's better to keep the previous logic if we don't want to break something. Nice catch!

I will do what you are asking, but let's go a little farther...

When my new approach cause test failure

As far as I know, there are no SunOS tests in the current CI. And, we both know, the green status is showing this.

So there should be some test cases outside the project, the dark matter, where people will have this issue.

When my new approach fixes problem where it could possibly cause test failure

AFAIK, SunOS could be configured to have the static zlib version in addition to the default dynamic one. In this case, keeping the old condition is a bug.

Let's ask the author of those lines @alanc

With my approach the user of this test case will be able to change his environment to adopt the usage on SunOS for his/her setup.

Symmetry / Special case

What is so special about SunOS to include it into this build script?

What about the default configuration on MS Windows (no Cygwin) or MacOS?

What about IBM i (AS/400) and BSD families?

Gentoo setup is configured for Boost with the help of SKIP_STATIC_BOOST in the GHA yaml file. Why could not SunOS be configured in the same way?

Bottom line

As I said, I will fix this to keep this special case for SunOS, but IMHO it must go.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

While there is no SunOS (Solaris or illumos) instance in the CI, we run the test suite on Solaris when building the meson packages for Solaris. It is true that it is not impossible for a user to build and install a static version of libz on Solaris - but since we provide a dynamic library for libz as part of the OS packages, I don't know of anyone who bothers to build their own static version. I can't speak to other OS'es or why they didn't hit the problem - perhaps their default configuration doesn't have dynamic libz either instead of just dynamic and not static.

I'm not opposed to moving this from a line of code to a configuration file - what's the "GHA yaml file"?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more. 

$ git --no-pager grep -C3 STATIC_BOOST
.github/workflows/cygwin.yml-        env:
.github/workflows/cygwin.yml-          # Cygwin's static boost installation is broken (some static library
.github/workflows/cygwin.yml-          # variants such as boost_thread are not present)
.github/workflows/cygwin.yml:          SKIP_STATIC_BOOST: 1
.github/workflows/cygwin.yml-        shell: C:\cygwin\bin\bash.exe --noprofile --norc -o igncr -eo pipefail '{0}'
.github/workflows/cygwin.yml-
.github/workflows/cygwin.yml-      - uses: actions/upload-artifact@v4
--
ci/ciimage/fedora/image.json-  "base_image": "fedora:latest",
ci/ciimage/fedora/image.json-  "env": {
ci/ciimage/fedora/image.json-    "CI":                "1",
ci/ciimage/fedora/image.json:    "SKIP_STATIC_BOOST": "1",
ci/ciimage/fedora/image.json-    "MESON_CI_JOBNAME": "linux-fedora-gcc"
ci/ciimage/fedora/image.json-  }
ci/ciimage/fedora/image.json-}
--
ci/ciimage/gentoo/image.json-  "env": {
ci/ciimage/gentoo/image.json-    "CI": "1",
ci/ciimage/gentoo/image.json-    "MESON_CI_JOBNAME": "linux-gentoo-gcc",
ci/ciimage/gentoo/image.json:    "SKIP_STATIC_BOOST": "1"
ci/ciimage/gentoo/image.json-  }
ci/ciimage/gentoo/image.json-}
--
ci/ciimage/opensuse/image.json-  "base_image": "opensuse/tumbleweed:latest",
ci/ciimage/opensuse/image.json-  "env": {
ci/ciimage/opensuse/image.json-    "CI":                  "1",
ci/ciimage/opensuse/image.json:    "SKIP_STATIC_BOOST":   "1",
ci/ciimage/opensuse/image.json-    "SINGLE_DUB_COMPILER": "1",
ci/ciimage/opensuse/image.json-    "MESON_CI_JOBNAME": "linux-opensuse-gcc"
ci/ciimage/opensuse/image.json-  }
--
test cases/frameworks/1 boost/test.json-  "matrix": {
test cases/frameworks/1 boost/test.json-    "options": {
test cases/frameworks/1 boost/test.json-      "static": [
test cases/frameworks/1 boost/test.json:        { "val": "true", "skip_on_env": [ "SKIP_STATIC_BOOST" ] },
test cases/frameworks/1 boost/test.json-        { "val": "false" }
test cases/frameworks/1 boost/test.json-      ],
test cases/frameworks/1 boost/test.json-      "b_vscrt": [

For all our docker-based CI testing we generate the docker images themselves with /ci/env_vars.sh as a kind of "/etc/profile for CI".

For Cygwin, which isn't docker-based, we use the GitHub Actions (GHA) yaml file .github/workflows/cygwin.yml. We don't maintain any such thing for Solaris' packaging.

Copy link
Author

@dememax dememax Nov 6, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you, @alanc !
@eli-schwartz, should we throw away the SunOS special condition, or merge the PR in the current state?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Excuse me, @eli-schwartz, what do you think?

@dememax
test cases/linuxlike/14 static dynamic linkage: Add matrix by static
9cec3d6 
Not only Solaris doesn't ship static libraries, it's common practice for
Linux distributions in general (security updates, size/space, etc.).
See, for example 'test cases/frameworks/1 boost'.

Enhancements:

  * Added flexibility: Introduced the SKIP_STATIC_ZLIB environment
    variable to optionally skip static linkage (similar to SKIP_STATIC_BOOST).
  * Expanded testing: Included testing for dynamic linkage, which was previously
    limited to static in verify_static.py (renamed to verify_zlib_linkage.py).
  * Improved test coverage: Ensured that each case is tested for
    its specific linkage type and the opposite case fails.
  * Addressed Cygwin issue: Ensured that the corresponding symbol is
    present on Cygwin.

Renames, comments, etc.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@alanc alanc alanc left review comments

@jpakkane jpakkane Awaiting requested review from jpakkane jpakkane is a code owner

@eli-schwartz eli-schwartz Awaiting requested review from eli-schwartz

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@dememax @eli-schwartz @alanc