"Burton, Ross" <[email protected]> writes: > On 30 March 2015 at 19:11, Richard Tollerton <[email protected]> wrote: > >> Running `false` interactively doesn't test this behavior because `sh -e` >> explicitly excludes interactive shells from consideration [1] [2]. >> You'll need to inject a `false` into run.do_terminal in order to >> reproduce this. >> > > I'm clearly not understanding how this behaviour is a problem then. What > broke?
What broke is devshell. If there's a bug in run.do_terminal that would cause a nonzero exit code during non-interactive execution -- e.g., http://cgit.openembedded.org/bitbake/commit/?id=18cd0ce6a55c9065c3f1bf223b47d817b5efcd8f -- then instead of a warning appearing in the devshell, the entire terminal quits a few milliseconds after it loads. In other words, this issue escalates other bugs that ought to be warnings into functionality-killers. This patch de-escalates them back to warnings. > Ross -- _______________________________________________ Openembedded-core mailing list [email protected] http://lists.openembedded.org/mailman/listinfo/openembedded-core
