[
Lists Home |
Date Index |
Thread Index
]
On 06/16/06 04:24, David Megginson wrote:
> That said, I think that part of the disconnect is between people who
> think in terms of serialized objects and people who think in terms of
> general information interchange, including tree models. There have
What you said in this and the next paragraph points at some of the
pitfalls of trying to exchange serialised objects across programming
languages. I presume you approve of interfaces that are built with the
intent of exchanging "documents", which I understand to be pieces of
information encoded in a well-defined representational structure,
without the attached notion of "types" or "identity", which are key
concepts in programming objects.
> JSON works well for AJAX because it's tightly targetted at JavaScript,
> just like Java RMI is tightly targetted at Java or DCOM is tightly
This analogy is correct but...
> targetted at Windows development. The other languages that have JSON
> libraries use them almost exclusively for exchanging information with
> Javascript in browsers, and as a result, developers are willing to
... I find JSON to be a pretty good "document exchange format" and
where I've used it we don't have a browser or JavaScript anywhere in
the picture. Essentially, it offers a structural representation of
scalars (with two fundamental types -- string, number), lists, and
maps (key-value pairs), which would be found in just about any
document representation format.
> live with the JavaScript style and idiosyncracies. Also, most of the
> JSON information exchanged is single-purpose (server-browser or
> browser-server messaging) and short lived (typically less than a
> second), so the design goals are different.
If I have to write an article I would still choose to write it in
something less intrusive like JSON than in XML. For now, I prefer to
hand write my reports in LaTEX ;-)
--
Tahir
|
