On 2022-03-02 10:52:30 [-0500], Steven Rostedt wrote: > On Wed, 2 Mar 2022 08:15:30 +0100 > Sebastian Andrzej Siewior <sebastian@xxxxxxxxxxxxx> wrote: > > > I was thinking about returning the error for compress/decompress via the > > return value and not touching errno at all. > > I know you hate the use of ERRNO in libraries and tooling, but we've decided > to go that way for all of trace-cmd and the libraries (libtraceevent and > libtracefs). This is just being consistent with all the other callers, and > to do it differently will create an inconsistency in the API. Ah okay then. In that case, the zstd interface does not set errno in the compress or decompress callback. This might need an update to be consistent. Otherwise the return value is < 0 and errno might still be 0. > -- Steve Sebastian
|