RE: [xliff] <bin-unit> problem

My apologies to Doug. It was Doug who added the discussion.

 

-jr

>>> John Reid <> 2/17/03 9:15:53 
PM >>>

In our discusion about <bin-unit>, which started with my proposal to 
add coord to <bin-unit> and was expanded by Bryan and Yves, it struck me 
that having a way to reference a <bin-unit> from a <trans-unit> 
would solve our problem. At first I thought we could do this by adding 
a bin-ref attribute to <x/> that takes the <bin-unit> id value as 
its value. 

 

Using the following example: 

 

<p>Click <img src="javascript:void(0);" alt="Start!" width="35" 
height="16"/> to start.</p>

We can store the width and height attributes in the coord of the 
<trans-unit> and move the image to a <bin-unit> that we'll 
reference:

 

<trans-unit id="11" 
coord="#;#;35;16">
  <source>Click <ph 
id='1'>&lt;img src="javascript:void(0);"image/gif" 
bin-ref="12"/></sub>" alt="<sub>Start!</sub>" 
/&gt;</ph> to start.</source>
  
<target></target>
</trans-unit>
<bin-unit id="12" 
mime-type="image/gif" reformat="yes">
    
<bin-source><external-file href="start.gif" 
/></bin-source>

</bin-unit>

This didn't work out so well for a few reasons:

1. coord in the <trans-unit id="11"> implies the entire text must fit 
in those coords, not just the image.

2. We still have to use a <sub> element for the image because an 
<x> cannot occur in a <ph>.

3. We have a reference id attribute, rid, used in the paired inline 
tags which is very similar to this bin-ref.

4. There is a <sub> for the alt text; thus, a segment within a 
segment.

 

Another way to handle this is to pull the entire <img> from the 
source. We do this by replacing bin-ref with rid in <x>.

 

<trans-unit id="11">
  <source>Click <x 
ctype="image" rid="12"/> to start.</source>
  
<target></target>
</trans-unit>
<trans-unit id="12" 
coord="#;#;35;16">
  <source><ph id='1'>&lt;img 
src="javascript:void(0);"image/gif" rid="13"/></sub>" 
alt="<sub>Start!</sub>" 
/&gt;</ph></source>
  
<target></target>
</trans-unit>
<bin-unit id="13" 
mime-type="image/gif" reformat="yes">
    
<bin-source><external-file href="start.gif" 
/></bin-source>

</bin-unit>

Now, when the coord is reformated upon 
localization, the new value is stored in the target of the <trans-unit 
id="12">. We still have the two <sub> segments within the one segment 
but this doesn't look to be a big problem since we've isolated the the entire 
image tag. We could go on step farther and add the rid attribute to <sub>. 
Then we'd have the following.

 

<trans-unit id="11">
  <source>Click <x 
ctype="image" rid="12"/> to start.</source>
  
<target></target>
</trans-unit>
<trans-unit id="12" 
coord="#;#;35;16">
  <source><ph id='1'>&lt;img 
src="javascript:void(0);"image/gif" rid="13"/>" 
alt="<sub>Start!</sub>" 
/&gt;</ph></source>
  
<target></target>
</trans-unit>
<bin-unit id="13" 
mime-type="image/gif" reformat="yes">
    
<bin-source><external-file href="start.gif" 
/></bin-source>

</bin-unit>

This could be pushed even farther by removing the alt segment with the same 
technique.

 

<trans-unit id="11">
  <source>Click <x 
ctype="image" rid="12"/> to start.</source>
  
<target></target>
</trans-unit>
<trans-unit id="12" 
coord="#;#;35;16">
  <source><ph id='1'>&lt;img 
src="javascript:void(0);"image/gif" rid="13"/>" alt="<sub rid="14"/>" 
/&gt;</ph></source>
  
<target></target>
</trans-unit>
<bin-unit id="13" 
mime-type="image/gif" reformat="yes">
    
<bin-source><external-file href="start.gif" 
/></bin-source>

</bin-unit>
<trans-unit id="14" 
>
  <source>Start!</source>
  
<target></target>
</trans-unit>

This is just thinking out loud. What do you think?

 

cheers,

john