[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [PATCH] -r { DATE } with words

From: Stefan Sperling <stsp_at_elego.de>
Date: Thu, 19 May 2011 11:53:18 +0200

On Wed, May 18, 2011 at 09:38:55PM +0200, Branko Čibej wrote:
> Why? That doesn't make sense. Second of all, all these wordy aliases are
> just shorthands for real timestamps anyway -- by your reasoning, you
> could eliminate all of them.

There is otherwise no way to express dates relative to the current time.
 
> But first of all, if you do have wordy aliases, they should be
> "compatible" with what people find in other tools (to wit, CVS) and what
> a speaker would find natural -- do you prefer "one day ago" to
> "yesterday" when speaking, even in German? :)

CVS had concepts in its date parser that don't even make sense in
the context of CVS (e.g. future dates) and had a ridiculous amount
of redundancy. That is why the old parser was dropped before 1.0.
I don't want to reintroduce that problem.

The initial commit was meant to include only the absolute minimum
of functionality needed to provide relative date stamps.
We can discuss adding more stuff, or removing this simple implementation
if you don't like it. However, it really is a useful feature for some
people. The person who pointed out this missing feature to me is now
planning to switch from CVS to SVN 1.7 because of this feature!
 
> By the way, one of the reasons why this was dropped before svn-1.0 is
> that it's almost impossible to translate.

All command line svn commands and their options are in English.
The date stamps are part of this interface so it is OK not to translate
them (and yes, doing so would be very, very hard).

> And now I see you parse "one
> days ago" which, I might add, is silly, let alone grammatically incorrect.

My original patch did not allow invalid grammer but Greg suggested
to accept it because it's easy to do and doesn't really hurt.
I'm indifferent.
 
> My take is -- either do this right, i.e., make it translatable and
> correct, or drop the feature altogether.

I think it's done right. There is no redundancy, it's very few lines
of code, and it solves the problem it is designed to solve (relative
dates). What isn't right about that?

If you want to be able to specify the time stamp "The Ides of March as
of the most recent year of the Chinese year of the dragon" (I stole
this quote) typed on the command line in ancient Egypt glyphs font,
then by all means feel free to make it work. I don't think that's the
right approach to this kind of feature. The language used should be
straightforward and simple but powerful enough to go back from the
current time by an arbitrary amount. That's what it is.
Received on 2011-05-19 12:01:43 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.