On Mon, 26 Mar 2007, C. Michael Pilato wrote:
> Erik Huelsmann wrote:
> > On 3/26/07, Daniel Rall <email@example.com> wrote:
> >> On Sun, 25 Mar 2007, Erik Huelsmann wrote:
> >> > In issue 2238, there's a comment about how the issue can be closed if
> >> > we don't resolve the issue by 1.5. Given the lack of interest in 'svn
> >> > propset/propdel <prop> <URL>', I propose closing the issue. (The
> >> > propedit case has been implemented and will be available in 1.5.)
> >> Does this mean that we support 'propedit' by URL, but not 'propset' or
> >> 'propdel' (on the grounds of slightly improved safety over the race
> >> condition)?
> > Currently, yes.
> You know, I used be in the group that felt that remote propset, propedit,
> and file modification was to be avoided because of the race condition. But
> the more I think about it, the more I wonder why I cared. After all, it's
> all versioned.
Non-rev-props, yeah. But we *already* let you set revprops by URL. I
don't understand the big deal.
> Would the race be less frightening if these actions required
> the user to pass --base-revision (the revision against which the changes are
> being made), and --base-revision was *not* allowed to have a value of
> "HEAD" (but instead had to be a real revision)?
Interesting sugeestion, hmm. It might be safer, but it would also
result in a more cumbersome user experience. I don't think that
there's a need for this.
Received on Tue Mar 27 02:49:23 2007
- application/pgp-signature attachment: stored