Re: [ogpx] (Not) serializing LLSD default values

Infinity Linden (Meadhbh Hamrick) wrote:
> i'm not absolutely sure all our legacy SL interfaces do this right 
> now, but i do know that we want them too. future VWRAP work will 
> definitely have those semantics.
>
> it may be advantageous to have that behavior be a configurable 
> parameter, then fire it up, and see what breaks. we could then use the 
> failure logs to track down interfaces that don't follow the "you get 
> the default if it's not there" semantics.
>
>

this might over-complicate things, but could it be possible to have 
contextual defaults?

Lawson

>
> On Wed, Dec 2, 2009 at 12:19, Hurliman, John <john.hurliman@intel.com 
> <mailto:john.hurliman@intel.com>> wrote:
>
>     I’m considering a change in the C# LLSD implementation
>     (OpenMetaverse.StructuredData) to skip serialization of default
>     values inside maps. For example, the following structure:
>
>      
>
>     {
>
>         “zero”:0,
>
>         “one”:1,
>
>         “empty_string”:””,
>
>         “hello_string”:”hello world”
>
>     }
>
>      
>
>     Would become:
>
>      
>
>     {
>
>         “one”:1,
>
>         “hello_string”,”hello world”
>
>     }
>
>      
>
>     From my reading of the draft (and our current implementation),
>     those two serializations should be equivalent. Do the current
>     Linden Lab implementations handle that correctly? The only problem
>     I see with the C# implementation is that it would make some
>     incompatible typecasts work when they shouldn’t. For example, the
>     receiving end could then read key “zero” as an LLSD array and it
>     would work, where the value “0” usually shouldn’t be convertible
>     to an array. So it assumes some level of coherence between the
>     sender and receiver, which I think is acceptable.
>
>      
>
>     I would look at the source of the other implementations myself,
>     but the last time I checked the PHP implementation was sorely out
>     of date and the C++ implementation was not under a liberal license.
>
>      
>
>     John
>
>
>     _______________________________________________
>     ogpx mailing list
>     ogpx@ietf.org <mailto:ogpx@ietf.org>
>     https://www.ietf.org/mailman/listinfo/ogpx
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> ogpx mailing list
> ogpx@ietf.org
> https://www.ietf.org/mailman/listinfo/ogpx
>   