Re: [Yaml-core] Tag defaults for ! and !! (#11c)

[Tim H. <tim...@co...>]

Clark C. Evans wrote:

>summary:
>  Ok.  This is an updated "patch" to #11 %TAG proposal, let's call
>  this #11c, it allows a "handle" to appear by itself, and in this
>  case an optional default specific is used.  Further this allows
>  !handle! to omit the exclamation marks within the %TAG proposal.
>  If the default is missing, it is the empty string.
>
>detail:
>  The %TAG directive has thre specific formats,
>
>    %TAG !           <prefix-of-no-handle>    <default-specific>?
>    %TAG !!          <prefix-of-empty-handle> <default-specific>?
>    %TAG !<handle>!  <prefix-of-handle>       <default-specific>?
>    %TAG <handle>    <prefix-of-handle>       <default-specific>?
>  
>
I'd ditch the alternate spelling <handle> and  insist on !<handle>!. The 
tag notation is pretty close to alphabet soup as is and allowing 
multiple spellings of the same thing, particularly when one of them 
(<handle>) is inconsistent with the rest is going to cause confusion.

I'm still going back and forth as to whether <default-specific> is 
needless complexity or important flexibility. Other than that, this may 
be (marginally) simpler than the alternatives while allowing everyone 
one enough rope to shoot themselves in their preferred appendage.

>  By default, there would be two %TAG directives in effect,
>
>    %TAG !   !                    
>    %TAG !!  tag:yaml.org,2002:   str
>  
>
Why bother? It seems simpler to just have ! and !! cook to themselves:

 %TAG !   !                    
 %TAG !!  !!

 Since the application controls all this anyway, it can make things work 
and we can move some complexity out of tagging and into the application 
/ schema. Since a 'standard' descriptions will need to supplied anyway, 
it can just as easily map '!' (or '!!') tagged items to string as it can 
map NULL tagged items to implicit. Call this #11C1

An even simpler is to have unspecified tags just map to themselves. This might hide errors, but presumably the application / schema will catch them. Call this #11C2.


example (adjusted for 11C1/2):

    ---
    # from proposal number #11
    private: !foo                   # !foo
    another: !foo::bar              # !foo::bar
    uri:     !tag:foo.com,2004:bar  # !tag:foo.com,2004:bar + warning
    yaml:    !!int                  # !!int
    handle:  !baz!bar               # -- error no %TAG for 'baz' (1) or !baz!bar (2)

    # from wart propoal #4
    nulltag:                        # NULL  (a special non-value)

    # default-tags
    also-private: !                 # !
    string:       !!                # !!
    empty-handle: !baz!             # -- error no %TAG for 'baz' (1) or !baz!
    ...

  If one were to provide a some explicit %TAGs

    %YAML 1.1
    %TAG !   tag:yatl.org,2002:                 wibble
    %TAG !!  tag:cla...@gm...,2004-08: hmm
    %TAG !baz! http://clarkevans.com/baz/
    ---
    # from proposal number #11
    private: !foo                   # tag:yatl.org,2002:foo
    another: !foo::bar              # tag:yatl.org,2002:foo::bar
    uri:     !tag:foo.com,2004:bar  # tag:yatl.org,2002:tag:foo.com,2004:bar
    yaml:    !!int                  # tag:cla...@gm...,2004-08:int
    handle:  !baz!bar               # http://clarkevans.com/baz/bar

    # from wart propoal #4
    nulltag:                        # NULL  (a special non-value)

    # default-tags
    also-private: !                 # tag:yaml.org,2002:wibble
    string:       !!                # tag:cla...@gm...,2004-08:hmm
    empty-handle: !baz!             # http://clarkevans.com/baz/
    ...

>notes:
>   This proposal allows but does not mandate that the '!' tag to be
>   loaded as what ever an application wishes, including a native
>   string.  But this, like the option to do regex-based typing based
>   on the NULL tag is completely under the application's control.
>   Of course, we could provide a 'standard' description of this
>   mechanism together with standard regex mappings to YAML tags
>   in a separate specification, and encourage implementations to
>   make this default behavior standard.
>  
>
To make the above work, the 'standard' mechanism would also want to 
treat '!' tagged scalars as  tag:yaml.org,2002:str And '!!x" tagged 
scalars as tag:yaml.org,2002:x. But all of this would be at the 
application/schema level. This is all just pushing complexity around, 
but I think that since YAML files want to be human editable, pushing 
complexity up to the application is good in this case.


-tim


>
>
>-------------------------------------------------------
>This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
>Project Admins to receive an Apple iPod Mini FREE for your judgement on
>who ports your project to Linux PPC the best. Sponsored by IBM. 
>Deadline: Sept. 13. Go here: http://sf.net/ppc_contest.php
>_______________________________________________
>Yaml-core mailing list
>Yam...@li...
>https://lists.sourceforge.net/lists/listinfo/yaml-core
>
>  
>