Johan Corveleyn wrote on Fri, Sep 21, 2012 at 14:52:59 +0200:
> On Fri, Sep 21, 2012 at 2:14 PM, Daniel Shahaf <danielsh_at_elego.de> wrote:
> > Johan Corveleyn wrote on Fri, Sep 21, 2012 at 11:46:21 +0200:
> >> On Fri, Sep 21, 2012 at 11:28 AM, Mat Booth <mat.booth_at_wandisco.com> wrote:
> >> > On 21 September 2012 10:20, Johan Corveleyn <jcorvel_at_gmail.com> wrote:
> >> >>
> >> >> ... a source of inspiration ...
> >> >
> >> > I see what you did there :-)
> >>
> >> Heh. To be clear: I haven't seen the code :-). But I've used the trial
> >> version of SmartSVN a couple of times to repair working copies at our
> >> company, so I can see the kinds of repairs that it does (in a much
> >> safer way than what I was doing myself (fiddling around with sqlite
> >> and trying various workaround with 'svn update -r0' and 'svn update
> >> --set-depth=exclude')).
> >
> > "In a much safer way"? How did it implement its repairs?
>
> I don't know how they implemented them, but at least they programmed
> the repair routines and (presumably) tested them. Which is already way
> better than me typing and potentially fat-fingering some 'sqlite
> .svn/wc.db "update nodes set ... where ..."' statements :-).
I wonder if SQLite databases have a snapshot capability? ("moment in
time" snapshots that are restoreable to --- like 'zfs snapshot' / 'zfs
rollback')
Received on 2012-09-21 15:36:24 CEST