Re: help in InkML verification

Hi Vinod,

"Vinod  Mankar" <v2k1@rediffmail.com> wrote:

> <traceFormat id="default">
>   <regularChannels>
>   <channel name="X" type="integer">
>   <channel name="Y" type="integer">
>   <channel name="T" type="integer">

"/" missing (closed elements)

>   </regularChannels>
> </traceFormat>
>

[...]

> wht i am trying is:
> 1. there are 3 channels, X, Y and the time channel T.
> 2. the values of the time channel are relative to the time specified
>   in each <trace> tags timeOffset attribute.
> 3. The timeOffset attribute in <trace> is recording the current time
>   of the day.

It looks to me like that's what your example is doing.

> wht needs to be clarified is tht
>
> 1. do i need to specify the <channelDef> for X and Y also ?
>   or they assume some defaults(wht)..

The <channelDef> definition is optional, even for channels that are found in <channel>
elements. If a <channelDef> doesn't exist, then the default value of the subelements
(representation, range, threshold, etc.) are assumed. Maybe the spec should be clearer
and add that statement, and add some missing default values. For instance I think that
the default for the type attribute of representation should be "decimal".

Maybe there should be default values for all the channels that have a predefined
name (see 3.1.4). Like B1 should always be boolean, with no units, or A should 
have a unit value of "radians". That way, files would only have to  <channelDef>
if the channels are non-standard. On the other hand, there might not be agreement
among devices what the defaults should be. Another drawback is that too many default
values makes files short but hard to read without the spec at hand. More feedback 
is welcome.

Max.

Received on Monday, 6 December 2004 17:17:17 UTC