On Feb 4, 2007, at 2:16 AM, Karl Fogel wrote:
> On 1/31/07, William Nagel <bill@stagelogic.com> wrote:
>> After years of participation in Subversion in a variety of
>> capacities, I've decided it's time to actually get involved in
>> Subversion development.
>
> Yay!
>
>> So, after taking a look at the various bite-
>> sized issues I decided to fix issue 2663 by adding a --non-recursive
>> option to svnlook. A patch is attached (with associated test) if
>> someone would be so kind as to take a look at it and critique.
>> Thanks!
>
> So, you picked a doozy this time -- because we're right in the middle
> of making the -N / --non-recursive flag obsolete, and replacing it
> with
> --depth=DESIRED_DEPTH instead :-). See
>
> http://svn.collab.net/repos/svn/branches/sparse-directories/
> README.branch
>
> for details.
>
> Now, note that for compatibility, -N will translate to --
> depth=files (that
> is, grab only files, don't get any subdirs). And I'm not actually
> against
> having a -N / --non-recursive flag appear on svnlook, since -N is a
> lot
> easier to type than --depth=files and it *is* the common case,
> after all.
> And the new depth-aware APIs being added on the branch don't
> affect your patch, since you're doing everything within svnlook/
> main.c,
> and using the indentation level to tell you whether you're on the
> zeroth
> level of the recursion or not :-).
>
> So, hmm, in retrospect, I guess the sparse-directories branch doesn't
> have anything to do with your patch, and I should rewrite this email.
> But I'm emotionally invested in it now, so I'll let it stand.
> Anyway...
>
> A comment on the code: you've made the 'indentation' parameter
> to print_tree() have a new meaning. The doc string for that function
> should change to reflect this.
>
> A comment on the behavior: I haven't tested it, but looking at the
> new body of print_tree(), it seems that your -N doesn't mean "print
> only files". Instead, it will actually print subdirectory names
> too, it
> just won't descend into those subdirectories and print *their*
> entries.
>
> This might be the most sensible behavior for svnlook, but it is
> inconsistent with -N's behavior elsewhere in Subversion. I think
> it might be good to raise a discussion on dev@ about the proper
> behavior of -N in svnlook, and then implement accordingly.
>
> Also, please provide a log message with your patch :-).
Thanks for the feedback, Karl. I'll tidy up the code based on the
points you made, and start up a discussion about how -N should work
for svnlook, then resubmit the patch.
-Bill
>
> -Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Feb 7 05:49:40 2007