On Thu, Feb 08, 2001 at 11:51:13AM -0600, Ben Collins-Sussman wrote:
> So we'll soon have two Repository Access (RA) implementations: ra_dav
> and ra_local.
> The client keeps available RA `plugins' (vtable structures) in a
> private array. When it sees an http:// repository URL, it knows to
> use the ra_dav vtable. When it sees a file:/ repository URL, it knows
> to use the ra_local vtable.
> My original idea was that libsvn_client load these RA implementation
> "on demand". But this is stupid, isn't it? I mean, libsvn_ra_dav and
> libsvn_ra_local should just be statically or dynamically linked, don't
> you think? (That is, rather than dealing with portably loading dso's)
> I can't see the advantage of on-demand loading unless we someday have
> 37 RA implementations...
> Do people agree?
The "plugin" concept and the *ability* (design-wise) to load them on demand
should be kept. But I believe we've been expecting to link them directly in
for a while now. We can deal with dynload later on.
But until then, keeping the interface narrowed through svn_ra_*_init() is
[ and don't forget the third: ra_xml! ]
Greg Stein, http://www.lyra.org/
Received on Sat Oct 21 14:36:21 2006