Re: [Libreoffice] [Libreoffice-qa] Java 7 support in LO 3.4.5 (was: minutes of tech. steering call ...)

On Fri, Dec 09, 2011 at 02:13:12PM -0600, Norbert Thiebaud wrote:
> On Fri, Dec 9, 2011 at 1:55 PM, Pedro Lino <pedlino@gmail.com> wrote:

> > > I know, I did it... but you don't have a 'push time'

> > :) Thank you, then :)
> > Why do I need to know the push time? Any commits that were pushed into
> > Central repository before time X are included in the source that is
> > pulled after time X... I think?

> sure. but then how do you known 'when' a given fix was pushed ? (and
> bear in mind timezone :-))

Oh come on, timezones is a solved problem: use UTC times.

As I was arguing on the other thread, commit time (as opposed to
author time) serves the purposes discussed rather well, no need to go
looking for push or pull time: because we usually keep a linear
history, it gives us a notion of "before" and "after". And even in the
cases we do a merge, the nodes in the graph that matter also have a
linear history: The nodes that matter are those that ever were the
HEAD of the branch. Even in case of a merge at M:


 A --- B ---- C ---- M --- D --- E
  \                 /
   T---F --- G --- H


The nodes T F G H were never the HEAD of the branch (e.g. master or
libreoffice-3-5), so never built by the (same) tinderbox, and M has a
commit time bigger than A, B, C and smaller than D, E.

Remember that the commit time is updated when one rebases or uses "git
am"; author time is not.

So, really, rather than "time at which the tinderbox pulled", I argue
that "recorded commit time of the HEAD node" is a better identifier to
put in tarball names, about boxes, etc. It is really (within a
branch) a proper global version number, à la SVN revision.

-- 
Lionel