On 08/12/2011 11:04 PM, Stefan Sperling wrote:
> On Fri, Aug 12, 2011 at 10:11:03PM +0200, Neels J Hofmeyr wrote:
>> Discussing 'svn info' output:
>> What do you think, for moved-here items, should we rename "Copied From URL"
>> and "Copied From Rev" to "Moved From *"? Leaving "Copied From *" as-is will
>> be confusing to the average user person...
>> #svn: "lol it says both copied and moved !?!1! Should be a tree conflict!1!"
>> Same discussion pending for --xml output.
> No, please let's not conflate copyfrom information with move information.
> Move information pertains to the working copy.
> It specifies how nodes were moved relative to the BASE tree.
> Copyfrom information pertains to the repository.
* neels looks up 'to conflate'
* neels looks up 'to pertain'
> I don't think there will be moved-from/moved-to info in the repository
> because this would require a fix for issue #898 (which I don't intend to fix).
> There will be copy-to information, and we'll use that together with
> copyfrom to figure out whether a node A was moved to B within the delta
> between two arbitrary revisions. This approach is like issue #3630.
> See this IRC discussion for information on my use of "move" vs. "true
> rename" and how all this relates to issues #898 and #3630/#3631:
This reads like the P.A.T.R.I.O.T. Act...
* neels indulges in a look-up frenzy
stsp, you are whipping out fancy words, issue numbers and mail threads mixed
with trivial info, about things entirely unrelated to *just* printing
"Moved" instead of "Copied" in 'svn' command line client output.
You said it, copyfrom belongs in the repository. Then why go ballistic about
whether the cmdline client prints "Copied From" or "Moved From"? A client's
local move has nothing to do with a committed copy! Indeed!
I'd be fine with a '-1', but I don't buy your (un-)reasoning.
>> We probably don't want to add a "Copied From Path" line: it's neither as
>> interesting nor as easy. A copy doesn't need to commit two paths in one, and
>> last time I checked we didn't store local paths for copied-from.
>> BTW, do we have any ambition to provide 'svn mv' using a URL source? :)
> No. Move information is only valid within the working copy.
Again, I get the impression that your mind has overlayed my question with an
unrelated concept you feel strongly about... But yeah, let's drop the 'svn
(though it *could* be a convenience thing that does a remote-delete + a
local-add-with-history, upon which you could've said "But that wouldn't be
committed in a single rev!", and you'd've been right about *that*.)
Received on 2011-08-13 03:01:53 CEST