On Wed, Feb 6, 2013 at 8:44 PM, Sean Silva <[email protected]> wrote:

> > I have a strong objection to coupling serialization to domain objects
> (that
> > usually explodes over time).
>
> Fair enough. IMO a free function is better too.
>
> > I'm for adding free-standing functions once we have a better idea how an
> > interface would look - but I'd like requests from other people to use it
> and
> > seeing their use cases.
>
> It might be interesting to just put something out there with basic
> functionality (and document it to make it discoverable), and see what
> people do with it (if anything). JSON output (even rudimentary output)
> opens the door to integration with lots of programming environments
> (and importantly, lots of _developers_).
>
> As you said, it's not that much code, so it doesn't really hurt to put
> it there on an experimental basis (but could lead to a big win). If
> you do this, please document it (as an experimental feature looking
> for users and which will be gladly enhanced when presented with use
> cases) somewhere in the introductory material to make it discoverable.
>

I stand by my argument based on the fact that it's really easy for
developers to output their own format to whatever they need.
I won't object to people throwing patches at me though ;)
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to