On Tue, 17 May 2005, Max Bowsher wrote:
> Peter N. Lundblad wrote:
> > On Mon, 16 May 2005, Greg Hudson wrote:
> >
> >> On Mon, 2005-05-16 at 15:10 -0500, lundblad@tigris.org wrote:
> >>> + Don't initialize libsvn_ra in svnversion:
> >>> + Justification:
> >>> + svnversion doesn't use the RA layer, and it only depends on it
> >>> + indirectly.
> >>> + Votes:
> >>> + +1: lundblad
> >>
> >> (1) You didn't include any justification in the log message for r14755.
> >> This justification is the only clue we have as to why you made that
> >> change.
> >>
> > You're right.
> >
> >
> >> (2) Who cares if svnversion only depends on the RA layer indirectly? We
> >> don't have any indirect functions for initializing it.
> >>
> > So you want svnversion to directly depend on libsvn_ra? I think the
> > current solution is OK for 1.2. We need to work htis out during 1.3.
> >
> > I'll fix the commit message.
>
> Wait, wait.
>
> svnversion does NOT depend on libsvn_ra.
> It depends on libsvn_client, which just happens to depend on libsvn_ra, but
> none of the functions that svnversion invokes in libsvn_client actually
> cause any use of libsvn_ra.
>
>
Yes, so from an API perspective it is wrong to call svn_ra_initialize from
a program that doesn't directly use libsvn_ra. But we should give an
application that uses libsvn_client to make sure libsvn_ra is initialized.
Let's work that out during 1.3. That was what I meant.
Regards,
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue May 17 15:13:08 2005