Re: [Yaml-core] survey regarding 23 vs '23'

[trans. (T. Onoma) <tra...@ru...>]

On Monday 13 September 2004 06:48 pm, David Hopwood wrote:
> Oren Ben-Kiki wrote:
> > Here's another option Clark didn't list, which I think sucks slightly
> > less than the others. I still think explicit-no-distinction is the
> > best, but this one also works for me, if I must:
> >
> > -
> >   name: default strings
> >   description:
> >     In this option, all non-plain scalars are given the default
> >     tag tag:yaml.org,2002:str. It is possible to force the parser
> >     to leave such nodes untagged by giving the "I have no tag"
> >     indicator (a single !). Plain scalars with no tag "have no tag"
> >     (have a NULL tag).
>
> +1 for explicit, +0.5 for this with Tim Hochberg's "! =3D NULL" suggestio=
n.

Have to argue against this one though. Here lets look at examples. Remeber=
=20
that NULL tags just mean that the app depends on the _kind_. (hence #2 may =
be=20
considered redundant)

#1
  ---
  - 23     # NULL
  - "23"   # NULL
  - []     # NULL
  - {}     # NULL

wart: requires ! to _prevent_ implicit typing of non-plain scalar

#2
  ---
  - 23     # ?var
  - "23"   # ?str
  - []     # ?seq
  - {}     # ?map

wart: requires ! on a ?str to get escaped plain-scalar (?var)

#3
  ---
  - 23     # !
  - "23"   # NULL
  - []     # NULL
  - {}     # NULL

wart: requires ! to _do_ implicit typing of non-plain scalar

#4
  ---
  - 23     # NULL
  - "23"   # tag:yaml.org,2002:str
  - []     # NULL
  - {}     # NULL

wart: non-plain scalars default to a global YAML type and
      require ! to make them "NULL" to _do_ implicit typing.


With #3 it seems to me like our wart has found it's way home. Since ! just=
=20
means implicit. It seems to have less downsides --at least not in the same=
=20
sense as the others. Its wart can also be thought of as a feature.

Very interesting to compare these. #1 and #2 are opposites in form, but #1 =
and=20
#3 are opposite in how the "wart" operates.

> -1 for having four kinds (there isn't sufficient justification to
> complicate the information model).

Agreed.

T.

=2D-=20
( o _  =E3=82=AB=E3=83=A9=E3=83=81
 //    trans.
/ \    tra...@ru...

I don't give a damn for a man that can only spell a word one way.
=2DMark Twain