| Summary: | Document precisely which toolchains and versions are supported for each version | ||
|---|---|---|---|
| Product: | [Documentation] General Docs | Reporter: | Laszlo Papp <lpapp> |
| Component: | docs-general | Assignee: | Beth Flanagan <elizabeth.flanagan> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Low | CC: | elizabeth.flanagan, ross.burton, sgw, srifenbark |
| Version: | unspecified | ||
| Target Milestone: | 1.6 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | 20 January 2014: IN PROGRESS REVIEW | ||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | Yes (doc changes required) | |
| Bug Depends on: | |||
| Bug Blocks: | 4941 | ||
|
Description
Laszlo Papp
2013-07-29 11:05:23 UTC
Some context as otherwise Scott will have no idea what this bug refers to. Currently busybox, udev, systemd (and likely more) fail to build when using a circa-2009 external toolchain, because their kernel headers are too old. Because we don't build our external toolchains we should document the minimum versions of the components they're providing. It is not restricted to that, really. The documentation should mention any distribution, toolchain, and so forth versions regardless my issue. That is just one of those, but I assume there is a lot more to it. So, do not get confused that is the only thing to fix for this issue to be resolved. Oh, and to avoid the confusion: I am not supporting arm-2009 to be deprecated. That is not the reason why I opened up this change before some people get the impression of that. That needs a different public discussion, I believe. *** Bug 4941 has been marked as a duplicate of this bug. *** Second reminder: don't change bugzilla state without reason. You did not just decide Scott's priorities. Actually, I was just about to put it to NEW, so more patience next time. It can be done in X + Y seconds even though if it is not done in X. :-) Also, I blame bugzilla because I wanted to reassign after the reopening, but bugzilla jumped to the next bugreport. This is so silly about bugzilla. All of the testing currently happens using the default internal toolchain, this toolchain defines the expected behaviour of any alternative toolchain that the user selects through TCMODE. Scott, can we add something like this to the reference docs for TCMODE: """ If TCMODE is set to a value other than "default" then it is your responsibility to ensure that the toolchain is compatible with the default toolchain. Specifically for Yocto 1.6 this is GCC 4.8, binutils 2.23, and linux-libc-headers 3.10. Using older or newer versions of these components may cause build problems. """ The versions will need to be verified at release time so if you have a list of sections that need to be updated, add this to it. Ross, I will add a note to the TCMODE that says the following: "If TCMODE is set to a value other than "default", then it is your responsibility to ensure that the toolchain is compatible with the default toolchain. Using older or newer versions of these components may cause build problems. See the Yocto Project Release Notes for the specific toolchain components with which the toolchain must be compatible." Then, in the Release notes we need to describe the situation again and list out GCC 4.8, binutils 2.23, and linux-libc-headers 3.10 specifically for release 1.6. After adding the note to TCMODE, I will re-assign to Beth. Scott I have updated the TCMODE variable to include the note. Please review at http://www.yoctoproject.org/docs/1.6/ref-manual/ref-manual.html#var-TCMODE. Scott Looks good to me. Re-assigning to Beth so that she is reminded to make a note about adding this to the future release notes. Not sure whether to turn the doc flag to "Done" or not. Probably not since it still requires Beth to note components in the release notes. I am not messing with the doc flag. Scott We are currently release noting this so I'm going to mark this resolved. |