-----Original Message-----
From: kfogel@newton.ch.collab.net [mailto:kfogel@newton.ch.collab.net]On
Would anyone object if we changed to HEAD:0 and made everyone's lives
easier? :-)
--------------------------
I think that changing "svn log"'s behavior would not be a good idea.
My first objection is the consistency one. As it is now, "svn log" does
what it "should" do, namely give you the history of the WC file, just as
"svn diff" gives you the difference between the WC file and its base. This
is not so important when users are typing the commands, and going to a more
"do what I (most likely) mean" definition could be an improvement. But as
soon as the commands are used in scripts, "do what I mean" starts to have
unpleasant results -- in the long run, commands that are intended to be used
inside scripts tend to become more and more rigidly and consistently
defined, often at the expense of ease of use by casual users, so that they
work consistently inside scripts. (Hence the cryptic output formats of all
the little Unix shell utilities -- they aren't really designed to be used by
people, but by shell scripts.)
And the "unknown future" problem can be a lot weirder than the example you
quoted. Consider creating file main/A/B/C in rev 100, copying it to
branch-1/A/B/C in rev 200, doing a bunch of edits to both main/A/B/C and
branch-1/A/B/C in revs 201 through 299, and then in rev 300, deleting
main/A/B/C and copying branch-1/A/B/C to main/A/B/C. In a WC with BASE at
rev 299, "svn log -rHEAD:0" tells *nothing* about the changes to main/A/B/C
in revs 201 to 299 and tells all the changes to branch-1/A/B/C -- because
that's the history of main/A/B/C at rev 300.
Dale
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Mon Jan 3 17:33:35 2005