Re: svn commit: r20677 - trunk/subversion/bindings/swig

Kouhei Sutou wrote:
> Hi,
>
> In <44BA6DA0.4070701@ukf.net>
> "Re: svn commit: r20677 - trunk/subversion/bindings/swig" on Sun, 16 Jul 2006 17:47:28 +0100,
> Max Bowsher <maxb1@ukf.net> wrote:
>
>> So, Kouhei, could you explain what advantages r19545 "Convert date C
>> string to Ruby Time in C level not Ruby level." brings?
>
> The advantage for me is that this conversion is natural for
> me. That is, converting a string that represents date in C
> to Ruby's Time is natural C <-> Ruby type conversion.

Agreed, but that was happening before r19545 anyway.

> And I think such type conversion is SWIG layer work.

This is the bit I'm asking about - why?

If I understand correctly, the change does have an efficiency benefit -
it avoids round-tripping a string from C -> Ruby -> C, but it does also
mean that the SWIG layer is no longer a direct wrapper of the C API -
which isn't intrinsically a bad thing, but if we do decide to do that, I
want it to be a concious decision, not something we move into accidentally.

> Anyway, if the project make a policy I'll follow the policy.

We have no policy yet, but I'm interested in making one.

Max.