Re: [xri] host-meta comments

[Dirk Balfanz]

On Wed, Feb 18, 2009 at 8:23 AM, Breno de Medeiros <breno@...> wrote:

While /host-meta is intended to be parsed by machines and not human-readable content, it is often the case that users eyeball such content for clues. For instance:

1. Developer is writing and debugging a library to parse host-meta files.
2. Developer is looking at /host-meta examples to get clues on how to write one for his site.

Being able to add human-readable comments on site-meta can be useful for such tasks. It also helps to preserve 'institutional memory' by documentation in place, which is often the only one that developers can locate.

Should there be a simple mechanism for line comments in site-meta?

+1 for comments.

I propose that any line that starts with # (possibly preceded by whitespace) is a comment.

Dirk.
 

--
--Breno

[phone] desk
[phone] (Grand Central)
MTV-41-3 : 383-A
PST (GMT-8) / PDT(GMT-7)

[Mark Nottingham-4]

The problem is that this will make it impossible to use an existing  
HTTP header parser (e.g., in Python, Perl, Ruby, whatever's standard  
library), a goal that's guided a lot of the design.

Why not just use

Link: </foo>; rel="something"
Comment: This one is for you, Joe!
Link </bar>; rel="joes-link"

?


On 19/02/2009, at 3:54 AM, Dirk Balfanz wrote:

>
>
> On Wed, Feb 18, 2009 at 8:23 AM, Breno de Medeiros  
> <breno@...> wrote:
> While /host-meta is intended to be parsed by machines and not human-
> readable content, it is often the case that users eyeball such  
> content for clues. For instance:
>
> 1. Developer is writing and debugging a library to parse host-meta  
> files.
> 2. Developer is looking at /host-meta examples to get clues on how  
> to write one for his site.
>
> Being able to add human-readable comments on site-meta can be useful  
> for such tasks. It also helps to preserve 'institutional memory' by  
> documentation in place, which is often the only one that developers  
> can locate.
>
> Should there be a simple mechanism for line comments in site-meta?
>
> +1 for comments.
>
> I propose that any line that starts with # (possibly preceded by  
> whitespace) is a comment.
>
> Dirk.

--
Mark Nottingham       mnot@...

[Breno de Medeiros]

Good point. Sounds good to me.

On Thu, Feb 19, 2009 at 4:00 PM, Mark Nottingham <mnot@...> wrote:

The problem is that this will make it impossible to use an existing HTTP header parser (e.g., in Python, Perl, Ruby, whatever's standard library), a goal that's guided a lot of the design.

Why not just use

Link: </foo>; rel="something"
Comment: This one is for you, Joe!
Link </bar>; rel="joes-link"

?

On 19/02/2009, at 3:54 AM, Dirk Balfanz wrote:

On Wed, Feb 18, 2009 at 8:23 AM, Breno de Medeiros <breno@...> wrote:
While /host-meta is intended to be parsed by machines and not human-readable content, it is often the case that users eyeball such content for clues. For instance:

1. Developer is writing and debugging a library to parse host-meta files.
2. Developer is looking at /host-meta examples to get clues on how to write one for his site.

Being able to add human-readable comments on site-meta can be useful for such tasks. It also helps to preserve 'institutional memory' by documentation in place, which is often the only one that developers can locate.

Should there be a simple mechanism for line comments in site-meta?

+1 for comments.

I propose that any line that starts with # (possibly preceded by whitespace) is a comment.

Dirk.

--

Mark Nottingham       mnot@...

--
--Breno

[phone] desk
[phone] (Grand Central)
MTV-41-3 : 383-A
PST (GMT-8) / PDT(GMT-7)