| Summary: | IMAGE_DEPENDS_* are added to do_rootfs[depends] | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Ola Nilsson <ola.x.nilsson> |
| Component: | Scripts and Tools | Assignee: | Ed Bartosh <eduard.bartosh> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium | CC: | ross.burton |
| Version: | 2.0.3 | ||
| Target Milestone: | 2.4 | ||
| Hardware: | All | ||
| OS: | Multiple | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | Yes (doc changes required) | |
|
Description
Ola Nilsson
2017-04-06 12:09:28 UTC
I did look and the code in image_types.bbclass does indeed put the dependencies onto do_rootfs rather than the specific image creation tasks. I think we want to move to a model where these tasks have explicit dependencies, e.g. instead of IMAGE_DEPENDS_jffs2 = "mtd-utils-native" we'd have do_image_jffs2[depends] += "mtd-utils-native:do_populate_sysroot". Since we're into feature freeze, I'm not sure how far we want to change things at this point in the release cycle and this may well need to be a 2.4 M1 activity. The end result may be the deprecation and removal of IMAGE_DEPENDS. I guess the rootfs dependencies are leftovers from the switch to the separated image tasks. I kind of like the way I can add just recipe names without the :do_populate_sysroot, but I can see the point in eliminating this set of variables. Anyway, it can certainly wait until 2.4. Replaced usage of IMAGE_DEPENDS with explicitly added dependencies to do_image_<image type> tasks: http://lists.openembedded.org/pipermail/openembedded-core/2017-June/138217.html Looks good to me. Surprisingly I manage to get to v4 so far :) http://lists.openembedded.org/pipermail/openembedded-core/2017-June/138386.html The fix has been accepted to poky master. Commit id: f93d58378fb4095a18c1403de03ca0e13ce465cd |