Hi,
   Thanks a lot for your advice Peter. We are actually in the case
which is described here: http://visualsvn.com/support/topic/00016/ We
are developing with C/C++. I agree, if multiple developers work
frequently on the same piece of code, there's probably a problem in
task assignment. Well, we are still a small project team, and not
fully organized. The second issue is that, we should be sure of that
we can apply both model with VisualSVN Server before fully settling a
system.
   I will share your email with my team mates.

Thanks again for your concern.
Best Regards

Caner

On Jan 5, 5:41 pm, "Bradley, Peter" <[email protected]> wrote:
> Hi,
>
> I'm sure you've carefully thought out what you're doing and have very
> good reasons for choosing a lock-modify-unlock model, but just in case
> no-one has brought this to your attention...
>
> A very large number of Subversion users would advise strongly against
> using a lock-modify-unlock system.  This is what Pragmatic Version
> Control by Mike Mason, 2nd edition, p26 has to say on the matter:
>
> "In reality, though, strict locking turns out to be a lot of extra
> hassle with no particular payback.  If you try an optimistic locking
> system (such as Subversion), you'll be surprised at just how rarely
> conflicts arise ... We've tried both kinds of locking over the years,
> and our strong recommendation is that the vast majority of teams should
> use a version control system with optimistic locking."
>
> Our experience is the same.  We introduced Subversion after years of
> struggling with Visual Source Safe.  VSS locks all checked out files and
> we were forever having problems with developers not being able to get to
> the files they needed to edit.  In the end we managed to persuade our
> management to go for svn with optimistic locking.  I can't tell you how
> much easier our life has been since then.
>
> Remember that you have to have two developers trying to edit the *same
> lines of code* at the same time for any conflict to arise.  If this
> happens regularly for you, the answer might be more easily found in
> project management practices than in source control configurations.
>
> If you've considered all this and still decided you need a
> lock-modify-unlock model, then please excuse and ignore me.  I just
> thought that someone ought to say it just in case you'd missed
> something.
>
> Cheers
>
> Peter
>

Reply via email to