CI: Test Against Geos Versions #362
Conversation
… factory test files
|
@BuonOmo I was able to get this to install and work across all versions on ubuntu systems but I'm having issues with cmake not working on some of the macos runners. It's strange because the issue is only happening on some of the runners. I can look into this more in a bit, but if you have any ideas/suggestions let me know. |
|
@keithdoggett by looking at the logs I can see that installing cmake installs version 3.27.2, but then while installing Geos we're looking for the version 3.27.3. Idk what software is asking for a specific version, but that is for sure the issue. Maybe some brew update may help. Might be brew upgrade. I'd have to look on my laptop. |
|
@BuonOmo good insight, thank you. Yes when I run |
|
@BuonOmo it's almost working now. For whatever reason the macos jruby builds can't find the libgeos_c library when loading it via FFI. I even tried specifying the path explicitly with an environment variable, but no luck. Not sure if there's some JRuby quirk I'm just not aware of that's causing the issue. Can investigate more in a little while. |
.github/workflows/CI.yml
Outdated
| - name: Set GEOS_LIBRARY_PATH (macos jruby) | ||
| if: matrix.os == 'macos' && matrix.ruby == 'jruby' | ||
| run: export GEOS_LIBRARY_PATH=/usr/local/lib/libgeos_c.dylib | ||
| - uses: ruby/setup-ruby@v1 |
There was a problem hiding this comment.
I tried on my mac, and I didn't need this GEOS_LIBRARY_PATH variable in my path.
$ ruby -v
jruby 9.4.3.0 (3.1.4) 2023-06-07 3086960792 OpenJDK 64-Bit Server VM 20.0.2 on 20.0.2 +jit [arm64-darwin]
$ geos-config --version
3.12.0
Although my geos installation is done through homebrew
There was a problem hiding this comment.
yeah I'll try again I'm pretty stumped as to why this it isn't recognizing Geos. I shouldn't have to specify a path but clearly something's not configured correctly.
|
@BuonOmo split out the CI between OS testing with package manger geos install and just ubuntu but with supported geos versions like we discussed. |
|
@keithdoggett thanks for getting that ball rolling ! It feels to me like a lot of redundancy both in execution (some tests are played twice) and in code. I don't have access to a laptop here to dig a bit the subject. Although I'm sure we could find a way to make this DRYer and still quite consistent and clear. What do you think? |
Summary
Picking up from where we left off on #355.