What YAML engine do you use?

A.M. Kuchling amk at amk.ca
Fri Jan 21 20:20:03 CET 2005


On Fri, 21 Jan 2005 18:54:50 +0100, 
	Fredrik Lundh <fredrik at pythonware.com> wrote:
> judging from http://yaml.org/spec/current.html (750k), the YAML designers are
> clearly insane.  that's the most absurd software specification I've ever seen.  they
> need help, not users.

IMHO that's a bit extreme.  Specifications are written to be detailed, so
consequently they're torture to read.  Seen the ReStructured Text spec
lately?

The basic idea -- a data dumping format that's human-readable -- isn't a bad
one.  OTOH, I can't recall wanting such a thing -- when I want readable
output I'm happy using
unreadable pickle files, unpickling the object and calling a .dump() or
.as_text() method.)

But YAML seems to have started out with the goal of being human-writable,
something you would write in Emacs, and that seems to have gotten lost; the
format is now just as complicated as Restructured Text, but more cryptic
(the URI namespacing for tags, for example), not really simpler than
XML and in some ways weaker (e.g. only two encodings supported, more
complicated escaping rules).

For a pure Python application, I can't see a need for YAML; use
pickle/cPickle instead, because they're already there.  Exchanging
serialized objects between Python/Perl/Ruby scripts might be a good use case
for YAML, but XML has wider software support and S-expressions are simpler,
so my inclination would be to use them instead of YAML.

--amk



More information about the Python-list mailing list