Re: Data Table Collections (Research)

2007-09-08 10:30:05 +0200 James Graham <jg307@cam.ac.uk>:
> Leif Halvard Silli wrote:
>> So why, have you included @SCOPE in your table example then? HTML4
>> doesn't need it. Only current HTML5 algorithm needs them.
> 
> If you turn off the "Use scope" option for the HTML 4 algorithm, the headers 
> are not quite right. In particular the "Day 2" cell and all cells below it 
> are incorrectly given the header "Day 1", and similarly the "Day 3" cell and 
> all cells below it are incorrectly given the headers "Day 1" and "Day 2".

I was unsure about whether it was right or wrong. But I now think now that you are correct that this is wrong.

But then the question comes (actually, it was Rob who said it, in a discussion I had with him): is this an accurate interpretaton of the HTML4 basic algorithm? That algorithm says «The search in a given direction stops when the edge of the table is reached or when a data cell is found after a header cell.» So why, then, does the Table Inspecor include Day 1 as header for the data cells under Day 2 - what is it based on?
 
> Each of these cells ("Day 1", "Day 2", etc.) is also given all the very top 
> row of heading cells as headers, again incorrectly. 

The same thing happens in the "experimental" and in the "smart col" algorithm, if the Table Inspector interprets those algorithms correctly.

I am not certain why it is incorrect. What else, should be correct? To not have any header cells at all? If I look for the header cells for e.g. Day 2, they it seems correct that I can hear or see all the header cells that applies for that Day.

> @scope is mainly used to  fix these issues.

>> Ben and James, are you also looking into improving the HTML4
>> algorithm? Or are you only looking to improve the HTML5 algorithm?
> 
> My interest is in specifying the best possibly algorithm in the final HTML 5 
> rec. Whether this looks functionally more similar to the HTML 4.01 algorithm 
> or to the current HTML 5 algorithm (which is clearly marked as unfinished in 
> the daft) depends only on which is the more useful.

I agree with the goal. 

But this is why I said what I said: from the Algorithm options that are available, it appears that the «Experimental» algorithm _includes_ the @headers attribute (as it is possible to select/unselect @headers for that algorithm), while the Smart Colspan algorithm does not offer to turn @headers on/off. Thus it seeems fair to say that the experimental algorithm currently tries to improve how things works in HTML4, with the @HEADERS and all. While the «Smart Colspan» tries to improve the drafted algorithm in HTML5.

I don't understand, though, what you mean by the «clearly unfinished» remark. I think that you ought to say in the table inspector which revision of HTML5 this is taken from - or find a more neutral name that can be used over a longer period. Because, we do not want to compare any of the other algorithms to a moving target.
-- 
leif halvard silli

Received on Saturday, 8 September 2007 14:15:13 UTC