Handle ragel better with meson #3383
Conversation
eli-schwartz
force-pushed
the
ragel-meson
branch
2 times, most recently
from
9c9fb91
to
2f55154
Compare
| @@ -16,7 +16,7 @@ jobs: | |||
| run: sudo apt-get install pkg-config gcc gcovr gtk-doc-tools libfreetype6-dev libglib2.0-dev libcairo2-dev libicu-dev libgraphite2-dev python3 python3-setuptools ninja-build gobject-introspection libgirepository1.0-dev | |||
| - run: sudo pip3 install fonttools meson==0.56.0 | |||
| - name: run | |||
| run: meson build -Db_coverage=true --auto-features=enabled -Dgraphite=enabled -Dchafa=disabled -Dragel_subproject=true -Doptimization=2 | |||
| run: meson build -Db_coverage=true --auto-features=enabled -Dgraphite=enabled -Dchafa=disabled -Dragel=disabled -Doptimization=2 | |||
There was a problem hiding this comment.
We no longer test the ragel subproject on ci builds, or am I missing something?
There was a problem hiding this comment.
The project options specify wrap_mode=nofallback
Do not run ragel unless the rl sources have actually been edited. If they have been edited, error with an informative error message
…oject() It does automatic subproject fallback and simplifies the code.
eli-schwartz
force-pushed
the
ragel-meson
branch
from
2f55154
to
66f009f
Compare
|
Is this ready to be merged? |
|
Why is the fedora bot failing? Should we instruct it to install ragel then? |
We shouldn’t need ragel on any CI jobs (unless we are specifically testing ragel generation, which i snot the case here). That being said, we do actually install ragel on this job but Fedora has the new ragel that we don’t want. This IMHO means the changes in this PR isn’t quite working yet because users shouldn’t need to do anything if they are simply building HarfBuzz and suitable ragel is not present. |
| ) | ||
| endforeach | ||
| message('You have to install ragel if you are going to develop HarfBuzz itself') | ||
| ragel = 'error' |
| endif | ||
| ragel_helper = find_program('gen-ragel-artifacts.py') | ||
| foreach rl : hb_base_ragel_sources |
There was a problem hiding this comment.
I think this part should remain in the else case of if not has_ragel, that's what make the CI fail AFAIK.
|
Is someone going to finish this? |
|
I lost track of this PR (sorry!) but I am still interested in getting it to work. |
Any chance you can finish this? |
Just like autotools, ragel should only be required when rl sources are actually out of date.
Fixes #3378