xmlhack: Evans moves against angle brackets in MinML

Clark Evans announced Yet Another Markup Language (YAML) to sml-dev, creating a markup language that has broken free of compatibility with XML syntax.

Acknowledging the paradox mentioned by Philip Nye:

The current Minimal XML spec is a subset of XML but is not compatible with many XML tools and extensions.

Evans has proposed to throw away the XML syntax:

Then let's break free. In our next "pass" let us _completely_ break from XML syntax. I'm serious here. I'm sick of tags and their connotations anyway!

After some more email exchanges, Evans created YAML, a curious mix of ideas borrowed from XML, but also from the syntax, variable types and declarations found in C, HTML, Perl, and Python and six fundamental requirements:

YAML texts are brief and readable.
YAML is very expressive and extensible.
YAML has a simple stream based interface.
YAML uses data structures native to your programming language.
YAML is easy to implement, perhaps too easy.
YAML has a solid information model, no exceptions no mess.

Evans believes YAML has enough qualities to have real potential and requests comments on the sml-dev mailing list:

If you haven't noticed... I'm pretty serious about YAML. I think it is a winner... it has a simple information model and a great syntax. I'm going to drive this forward, and I'd like your thoughts... what are your requirements? Thus far YAML is a information serialization mechanism.

Other stories:

Sure, I'll use a non-standard mark-up language that has zero industry support. Good one. Very convincing.

Re: Evans moves against angle brackets in MinML (Clark Evans - 03:30, 16 May 2001)

(comment on my comment) I did not intend to be this harsh about URIs. Clearly they are a very useful mechanism, they just bring complexity that I am not convinced YAML should meddle with. YAML will have a reference mechanism (aka href/id) and perhaps this may be extend to allow any arbitrary URI, but this is a big jump in complexity that I'm not sure YAML should take. Also, we may be adding classes and for this some reasonable namespace mechanism will be required. I am not sure what it will be but I can assure you we will steer clear of prefix abbreviation mechanisms and relative uri references. Thank you so much for your interest and I hope to see you at http://www.yaml.org.

Kind Regards, Clark Evans

Re: Evans moves against angle brackets in MinML (Dan Brickley - 06:27, 14 May 2001)

Interesting... though without a namespaces mechanism, it seems a bit toy. The information model appears to map into RDF's edge-labelled graph structures rather well, were it not for the lack of URIs on the edges/properties ('invoice','date','buyer' etc) and the constraint that multiple edges with the same name aren't allowed. I suspect that if these tweaks were added, it would start to look rather like N3 (http://www.w3.org/2000/10/swap/Primer.html). The N3 examples at http://www.w3.org/2000/10/swap/Examples.html make the case for using URIs in the information model, to facilitate data-merging. I'd like to have a play with a Perl implementation...

--danbri