On 09/10/2013 04:59 PM, Michael Matz wrote:
Hi,
On Tue, 10 Sep 2013, Adam Spiers wrote:
The timestamp is useless as mentioned previously
Mentioned by who? It's certainly not useless - it's *required* for correct chronological ordering, as I already mentioned.
You're too fixated on git. Correct chronological ordering is not depending on timestamps. It depends on some number that monotonically increases with commit time. svn revision numbers do for instance.
And as timestamps indeed have the mentioned problem it's actually a bad way to ensure chronological ordering. Better than nothing I suppose, but not very good.
True for svn (and similar ones). Since branches are only directories here and the history is always global, revision numbers should be sufficient. However, commit timestamps should equally comply. So using both could be viewed as overkill / redundancy, but at least it wouldn't hurt in some general packaging recommendation. If possible, I'd want to avoid documenting many special cases since that would somewhat defeat the purpose of this thread. Being aware of things is certainly useful, so thanks for broadening the discussion.
An offset (with or without hash, depending on what's needed) is both short and precise.
Sorry, I don't understand this. Can you explain in more detail?
Offset == number of commits on top of some base (where base in git describe parlance is e.g. a tag).
Ciao, Michael.
-- Sascha Peilicke SUSE Linux GmbH, Maxfeldstr. 5, D-90409 Nuernberg, Germany GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer HRB 16746 (AG Nürnberg)