Ben Collins-Sussman wrote:
> I'm unclear as to what Hyrum's motivation is. If the motivation is to
> make the svn commandline client 1.65MB instead of 1.7MB, then I think
> this is not worth our energy. If the motivation is to get rid of
> compile-time warnings, then it's worth a bit of energy -- we've always
> prided ourselves on having no warnings. (Heck, people build svn source
> in *live public presentations* because it's famous for having no build
> warnings!)
>
> But I agree with gstein: if this means increasing the risk of
> 'accidentally incompatible' libraries in any way, then I think this is
> a dangerous path.
Decreasing the amount of compile-time warnings is one goal. Currently, the only
place we use deprecated functions is in the tests and other deprecated functions
themselves. The other reason is to allow projects such as TortoiseSVN to ship a
binary with less dead code than they otherwise would.
But it's not a big deal. The work's already been done, partly just because I
was curious what portion of our binary was backward compatibility code. I'm
happy to shelve it if desired.
-Hyrum
Received on 2008-09-30 16:26:40 CEST