So, to summarize, the motivations are
* doesn't require perl/python on Windows
* doesn't require third party scripts / extra packages on Unix
* useable by API clients
* canonical implementation
Stefan Sperling wrote on Tue, Jun 21, 2011 at 12:03:22 +0200:
> On Tue, Jun 21, 2011 at 12:55:15PM +0300, Daniel Shahaf wrote:
> > Stefan Sperling wrote on Tue, Jun 21, 2011 at 11:34:46 +0200:
> > > On Tue, Jun 21, 2011 at 12:21:27PM +0300, Daniel Shahaf wrote:
> > > > Please convince me why having svn_client_bisect() is needed.
> > > >
> > > > There are already THREE scripts out there that implement cmdline 'svn bisect':
> > > > one on CPAN, one in Peter's Debian package, one that Julian just posted.
> > > > (And we got one sent to the issue tracker a while ago, but the CPAN one
> > > > was more featureful)
> > >
> > > Julian's implementation is, by his own account, incomplete.
> > >
> > > There is precedence here. Let me provide two examples:
> > > We have svnmerge.py. Why is there merge-tracking in svn now?
> > > We have 'svn patch' now. Wasn't UNIX patch good enough?
> > >
> >
> > No, because it won't handle properties and renames.
>
> Another primary reason was that Windows users now have a patch
> feature withouth installting third party software.
>
> > What value does
> > having 'svn bisect' in the core add over the existing external
> > 'svn-bisect' implementations?
>
> It's handy for Windows users or even Unixy types who don't want
> to install zillions of helper scripts.
>
> I am not even going to consider ever trying to fix a bug in an
> svn-bisect implementation that was written in Perl. git-svn was
> enough experience of that, thank you very much.
>
> I suppose Arwin could share his ideas in detail once they have
> developed, and explain advantages over the existing scripts.
Received on 2011-06-21 12:26:54 CEST