Hi On Wed, Aug 23, 2023 at 4:31 AM Stephen Brennan <stephen.s.bren...@oracle.com> wrote: > > Stephen Brennan <stephen.s.bren...@oracle.com> writes: > > Marc-André Lureau <marcandre.lur...@gmail.com> writes: > >> I am a bit reluctant to change the dump format by default. But since the > >> flatten format is more "complicated" than the "normal" format, perhaps we > >> can assume all users will handle it. > >> > >> The change is probably late for 8.1 though.. > > > > Thank you for your review and testing! > > > > I totally understand the concern about making the change by default. I > > do believe that nobody should notice too much because the normal format > > should be easier to work with, and more broadly compatible. I don't know > > of any tool which deals with the flattened format that can't handle the > > normal format, except for "makedumpfile -R" itself. > > > > If it's a blocker, I can go ahead and add a new flag to the command to > > enable the behavior. However there are a few good justifications not to > > add a new flag. I think it's going to be difficult to explain the > > difference between the two formats in documentation, as the > > implementation of the formats is a bit "into the weeds". The libvirt API > > also specifies each format separately (kdump-zlib, kdump-lzo, > > kdump-snappy) and so adding several new options there would be > > unfortunate for end-users as well. > > > > At the end of the day, it's your judgment call, and I'll implement it > > how you prefer. > > I just wanted to check back on this to clarify the next step. Are you > satisfied with this and waiting to apply it until the right time? Or > would you prefer me to send a new version making this opt-in? >
Nobody seems to raise concerns. If it would be just me, I would change it. But we should rather be safe, so let's make it this opt-in please. -- Marc-André Lureau