? at beginning of prefixed field

[Olly Betts]

On Sat, Mar 14, 2026 at 07:12:40AM +0900, David Bremner wrote:
> I guess you saw the build is failing on the autobuilders. I duplicated
> that failure in sbuild on my local machine. On the other hand, if I
> disable the tests, the result does seem to fix the problem I reported,

Yes, I saw but hadn't had a chance to work out what was going on.

The new tests do pass in (upstream) git master.

When I built the package, my machine had a load spike and a random
testcase timed out under the remote backend.  That seemed to be the
only failure, and I knew the tests passed upstream, so I cut a corner
and to avoid having to wait for all the tests to run again I built the
package without running the tests and uploaded.  I must have missed that
the new testcase had also failed.

It looks like the issue was that the generated parser code isn't being
regenerated during the package build (I thought it was, but it isn't
in the package build log).  Though that would mean your result of
building with the tests disabled shouldn't actually contain the fix...

It'd probably be better if we did regenerate the parser during the
package build, but this close to a new release I'm favouring minimal
changes so I've added the generated file to the patch, rebuilt the
package (test pass locally) and uploaded.

Cheers,
    Olly