On Fri, May 20, 2011 at 04:28:55PM +0200, Neels J Hofmeyr wrote: > On 05/18/2011 09:38 PM, Branko Čibej wrote: > >On 17.05.2011 11:36, Stefan Sperling wrote: > >>On Tue, May 17, 2011 at 12:45:50AM +0200, Stefan Sperling wrote: > >>>Any comments or objections? > >>Neels didn't like the arbitrary "round to 00:00 of next day" rules > >>and everyone in the hackathon room seems to agree. So "one day ago" > >>is now the same as "24 hours ago". > >> > >>I also dropped the "yesterday" keyword because it overlaps with "one day > >>ago". > > I liked 'yesterday' as 'yesterday' == 'one day ago', even add > 'fortnight' and so on. But I agree with Brane's reservations. Having > an untranslatable, grammar dependent non-feature that isn't even > documented... weeeell... > > BUT, why don't we just use standardized unit letters? e.g. {-1d} > means one day ago. Then we'd have something like > > [-+]<float-nr>[YyMDdHhmSs]
What would we need the + for? We cannot resolve future revisions. > (M = months, m = minutes, the others are case insensitive) > No need to translate, no grammar involved, less characters to write. > > svn log -r {-1.5d} > (one-and-a-half days ago) > > Heh and we'd drop the '+' too, I guess, making the '-' optional. > Brainstorming more, the [+-] could define the range: > > svn log -r {1y} -r {+1M} > > (get the log from one year ago to a month after that.) > and the reverse case, though some may probably not want to support this: > > svn log -r {-1d} -r {5M} > svn log -r {5M} -r {-1d} > > (get the log from five months and a day ago until five months ago.) > > If I were to write this feature, this would be my choice. stsp? I think the current simple grammer is easier to understand at first sight. It doesn't even need documentation. -r{-1d} isn't self-documenting at all.