[stringtemplate-interest] Odd even row formatting

[Terence Parr]

Hi guys,Sorry for delays in responding to mail what I'm racing to get  
antlr v3.1 finished and grade my final exams... answers below. good  
questions!!

On May 15, 2008, at 4:09 PM, Trevor Griffiths wrote:

> Hi,
>
> The topic is how to change the style of rows in a table based on  
> their odd or even index value.
> Or more generally, given a set of ordered data to iterate over how  
> to make the style dependent upon the data's position in the the set.  
> That is on the index of the data.
>
> The current ST suggestion (at least in the  documentation) is to set  
> up this kind of template structure:
>
> $subjectList: oddRow(), evenRow()$
>
> Where the oddRow template might be:
>
> <tr>
>     <b><td>$it.subject$</td><td>$it.count$</td></b>
> </tr>
>
> And the evenRow template might be:
>
> <tr bgcolor= red>
>     <td>$it.subject$</td><td>$it.count$</td>
> </tr>
>
> The subjectList is iterated over and every even row will have a red  
> background.
>
> One problem I have with this approach is that it seems to violate  
> the DRY principle.

Well, as you say, this one is pretty simple, but the principle is  
correct. There will be repetition. In that case I suggest that you  
simply use another template to factor out there, and elements like

row2col(left,right,color) ::= ...

then do

oddRow() ::= "<row2col(a=it.subject,b=it.count)>"
evenRow() ::= << <row2col(a=it.subject,b=it.count, color=red)> >>

> From a separation of concerns perspective, is there any reason why I  
> shouldn't be able to explicitly key off the index, either through an  
> odd/even boolean of some sort? This would allow this kind of  
> expression:
>
> <tr $if (odd)$ bgcolor= red $endif$ >
>     <td>$it.subject$</td><td>$it.count$</td>
> </tr>

>
> And while odd/even is a fairly common requirement, it begs the  
> general question of why can't I change the presentation of the data  
> based on its index value? I'm not attempting to change the model. I  
> am simply trying to: "operate on a set of incoming read-only data  
> values, called attributes..."

switch on iteration is something that is fine; the only problem is  
finding a construct that allows only that and not generic testing of  
integers etc. I wonder if something could be done with a map and using  
the index as keys or something.

> Actually, I did just implement odd/even in the model by computing  
> odd/even as each subject line bean is loaded in to an array of beans.

also, can use implicitly created aggregates

st.setAttribute("aRow.{odd,data}", false, mydataobject);

  this makes it much easier then you can simply test $it.odd$ and use  
$it.data$ to get your data element.

> But, now don't  we have a case of the model entangled with the view?  
> there seems something slightly awry here or is it just me :-)

Not really in that the model can know whether something has an even or  
odd value ;)

> So, in conclusion, Is it reasonable to have an odd/even attribute,  
> if not the index, available in the presentation layer, or is this  
> the first step on a slippery slope :-)

Perhaps a switch or something I was limited to the iteration?

> If the index is made available, then it would obviously need to be  
> compared with some value, which breaks the simplicity of the "does  
> it exist" rule.

yeah, we would have to hide it so did not look like a comparison ;)

Ter