Re: [OE-core] [PATCH 1/3] oeqa/sdk: run the SDK's own copy of a host tool

[Richard Purdie <richard.purdie@linuxfoundation.org>]

On Sun, 2026-08-30 at 14:36 -0400, Trevor Woerner via lists.openembedded.org wrote:
> ensure_host_package() decides whether a test can run by looking for a
> package name in the SDK's host manifest, which establishes neither that
> the package is present nor that the command the test runs is the SDK's
> own. On an extensible SDK the manifest lists what the shipped sstate
> could provide, so a case can pass the check and then run a command that
> is not there, or one belonging to the build host.
> 
> Ask the SDK for the command instead. An eSDK ships sstate for more than
> it installs, so when the command is missing devtool installs the recipe
> and the lookup runs again. A command that resolves outside the SDK now
> fails rather than reporting a result about the build host.
> 
> AI-Generated: codex/claude-opus 5 (xhigh)
> Signed-off-by: Trevor Woerner <twoerner@gmail.com>
> ---
>  meta/lib/oeqa/sdk/case.py | 96 ++++++++++++++++++++++++++++-----------
>  1 file changed, 69 insertions(+), 27 deletions(-)

I'll be honest, this feels/looks wrong. In order to say why, I need to
spend a lot of time looking into the problem more, trying to work out
the right way to solve it and so on.

If I keep doing this, people will just keep feeding me patches like
this and expecting me to give that level of feedback which I simply
can't do, it isn't easy.

So no, this isn't right and no, I don't feel it appropriate to merge
it, sorry.

Cheers,

Richard