Feature/jazzy#558
Conversation
|
I'm hitting but I'm not sure why yet.. Copilot says both the cpp_generator_aspect and c_generator_aspect are being applied to the same targets simultaneously, trying to write identical output files because we explicitly define interface libraries like this: I have doubts there, but it surely is related to the |
|
here's another issue: |
|
Something like this will work to fix the diff --git ros2/interfaces.bzl ros2/interfaces.bzl
index 1862319..f270df1 100644
--- ros2/interfaces.bzl
+++ ros2/interfaces.bzl
@@ -280,7 +280,7 @@ def run_generator(
generator_templates = generator_templates[DefaultInfo].files.to_list()
generator_arguments_file = ctx.actions.declare_file(
- "{}/{}_args.json".format(package_name, generator.basename),
+ "{}/{}{}_args.json".format(package_name, generator.basename, "_" + mnemonic if mnemonic else ""),
)
output_dir = generator_arguments_file.dirname
type_description_tuples = type_description.type_description_tuples if type_description != None else [] |
|
here's the next issue: |
|
I just expunged cache and |
|
@lalten In case you have custom aspects that build on top of aspects from rules_ros2. Here's the new diagram: flowchart
idl_adapter_aspect --> type_description_aspect
type_description_aspect --> c_generator_aspect
c_generator_aspect --> cpp_generator_aspect
c_generator_aspect --> py_generator_aspect
i.e. there is a new aspect type_description_aspect that c-generator depends on. In addition cpp-codegen depends on c-codegen. Duplicates that could happen during codegen are fixed in repositories/patches/rosidl_fix_rosidl_typesupport_introspection_generator.patch and repositories/patches/rosidl_typesupport_generate_true_c_code.patch. |
|
@lalten Any chance you'd have a bit more time sometime soon to take a look again at this PR? |
Sorry, it's unlikely I'll find time until mid jan. I'd be ok with landing this and then figuring it out later :) |
There was a problem hiding this comment.
I think I got the Jazzy upgrade to work but I would need a bunch of changes in this (and ament.bzl) due to IDL support patches that we've not upstreamed yet.
@mvukov would it be possible to decouple these changes from the Jazzy upgrade and land them first?
There was a problem hiding this comment.
Sure thing. I'm not in rush -- I'd just like to merge it sometime soon.
|
I updated one repo to use this branch, things seem to be working fine. I think it'd be good to merge this to main and fix any issues that we may find later. Also I think maintaining Humble is dispensable. While it would be nice to have, it's probably not worth the effort. Providing a |
|
Agree with ahans. Python 3.10 is starting to hold other things back for us. |
No description provided.