John Beck wrote: > Darren> At the close of an onnv milestone build there is an hg tag added, > Darren> eg onnv_118. What happens if a changeset needs to be backed out > Darren> and a respin done to ensure that the build is usable ? I see that > Darren> recently there was such a case and backout push was done but the tag > Darren> for onnv_118 is still set before the backout. > > Darren> While in this case the divergence between the onnv_118 tag changeset > Darren> and the backout one isn't much (only two changesets) it could in > Darren> theory be more. > > Darren> So if projects want to be in sync with build close and a backout > Darren> happens post build close but is required for the stability what to > Darren> the ON gk's recommend we do ? Or is it so rare as not to be worth > Darren> really worrying about that much ? > > JBeck> ... good question; let us (the ON gk's) discuss > JBeck> it amongst ourselves, then one of us will respond later today ... > > We agreed that henceforth we will send a heads-up message to the usual > lists when a significant respin occurs. Note that not all respins are > equal; some are to correct little things like package inconsistencies. > But for "big" things (which this case definitely was) we will announce > the details of the respin. In fact, James is working on the (belated) > announcement for this one right now. > > And we will resume the practice that inadvertently got dropped when Steve > Lau left of generating a Mercurial bundle file containing any changes > made after the build tag was created, and publish it as before at: > > http://dlc.sun.com/osol/on/downloads/hg-build-snapshots/
Fab! That is IMO even better than a tag because it makes it really really obvious what is going on. The one tiny downside it has (that branches might be able to solve) is that all the history isn't in one place but I can live with that - after all it only applies to intermediated biweekly builds anyway. -- Darren J Moffat
