W3C home > Mailing lists > Public > xproc-dev@w3.org > April 2009

Re: Compute attribute value with set-attribute

From: mozer <xmlizer@gmail.com>
Date: Tue, 7 Apr 2009 20:55:38 +0200
Message-ID: <21d9ade60904071155n290abf70sfe8ac911cd30b910@mail.gmail.com>
To: Florent Georges <fgeorges@gmail.com>
Cc: James Sulak <jsulak@gmail.com>, XProc Dev <xproc-dev@w3.org>
On Tue, Apr 7, 2009 at 6:02 PM, Florent Georges <fgeorges@gmail.com> wrote:
> 2009/4/7 James Sulak wrote:
>
>> This is just an idea; I'm not sure if it's actually a good one.
>> In practice, I doubt this problem will happen much, since steps
>> generally have a default readable port available.  And the
>> current approach has the virtue of consistency, even if it is
>> more verbose.
>
>  I am a bit concerned about that also, but the other way
> around :-) I wonder if this is really intuitive to implicitly
> bind a step result to a p:with-option.  In my humble opinion, it
> is intuitive to connect a step's output to the next step's input.
> This is less obvious that a step's output is connected to the
> following step's with-options that are not explicitly bound to
> any port.
>
>  I do not mind to add a p:empty (well, if I know I have to :-/)
> but I wonder if this is really helpful to implicitly bind
> p:with-option to the default readable port, instead of forcing
> the user to explicitly bind it if she wants to.
>
>  I guess that will give a processor less opportunities to detect
> user's errors while this does not really improve readibility or
> conciseness.  In the following example:
>
>    <p:identity name="you">
>       <p:input port="source">
>          <p:inline>
>             <what>you</what>
>          </p:inline>
>       </p:input>
>    </p:identity>
>
>    <p:add-attribute match="elem" attribute-name="a">
>       <p:input port="source">
>          <p:inline>
>             <what>world</what>
>          </p:inline>
>       </p:input>
>       <p:with-option name="attribute-value" select="
>           concat('Hello, ', what, '!')"/>
>    </p:add-attribute>
>
> I am not sure this is really more readable than:
>
>    <p:with-option name="attribute-value" select="
>        concat('Hello, ', what, '!')">
>       <p:pipe step="you" port="result"/>
>    </p:with-option>
>
> and in the later I am sure the user intended to bind the
> with-option to the result of the step "you."
>
>  Well, just my humble opinion.

I agree with your analysis, but you miss one important point in it :
you may end up having to give a name to a bunch of component for that,
which might become troublesome at some point

I agree with James S. : it is consistant but not ideal

It reminds me the double quoting problem of strings in XPath/XSLT

Xmlizer
Received on Tuesday, 7 April 2009 18:56:18 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 7 April 2009 18:56:18 GMT