Replacing std.xml

[H. S. Teoh]

On Thu, Aug 29, 2013 at 01:14:19PM +0200, Chris wrote:
> On Thursday, 29 August 2013 at 09:24:31 UTC, Joakim wrote:
> >I think it's great that there's no std.xml, as it implies that
> >nobody using D would use a dumb tech like XML.  Let's keep it that
> >way. :)
> 
> No way around XML. A must have, as has been said in this thread. But
> what would you suggest as a better alternative to XML. It might be
> worth creating modules for alternative too (like JSON).

While I do agree that in the current state of affairs, XML support is a
must, I also think that XML is just way overengineered, IMNSHO. It has
adds too much overhead and therefore requires compression to be
efficient, and it is needlessly complex for what it does (tag
attributes, all the different cases of CDATA / non-CDATA, etc.). This
complexity makes it impractical to edit by hand, relegating it to
machine reading/writing only, which then begs the question of why a
binary format wasn't chosen instead. And don't get me started on DTDs,
which are incredibly convoluted and can't even express certain things
that one might want to express in an automatic validation system. Or
that 17-headed monster called XSLT, which, thankfully, is fading into
the obscurity of time.

JSON is a nicer, simpler alternative, though there may be limitations
with it that I don't know about. Word on the street is that many people
are abandoning XML for JSON due to lower maintenance overhead (and this
includes one of my friends, who was a hardcore XML fanatic -- I was
frankly quite surprised when he told me he was considering migrating to
JSON, since the original reason he chose XML was so that his data will
future-proof... well, so much for *that*).

But all of this is irrelevant... it doesn't alleviate the need for a
std.xml replacement, since we have to live in the real world where XML
exists and must be supported. :)


T

-- 
Life would be easier if I had the source code. -- YHL