Proposed table specification (long!)

[Michel Fortin]

Le 2011-05-11 à 9:41, Simon Bull a écrit :


> Thanks for your comments Michel.


You're welcome.


> In reply to the points you raise:

> 

> 

> Regarding complexity:

> It is not clear to me whether folks are objecting to _parsing_ complexity or

> *reading/writing* complexity. Subjectively I don't think the example is

> difficult to read; it couldn't be much simpler.  So I will assume that

> people are concerned about parsing complexity.


It's pretty easy to read, no complexity problem there. I think the complexity lies in parsing, writing/editing, and explaining/understanding the possibilities.


> Implementation considerations should not drive the formulation of the specification except

> where some absolute technical limitation dictates otherwise.


True, up to a point. It isn't worth investing tons of your time for a feature that'll benefit very few people (unless maybe yourself are one of those people, of course).


> A markdown document should be *publishable* _as-is_.  Wobbly mis-aligned

> tables do not make publishable documents in any profession as far as I know.


Well, the introduction says that indeed. Except that the role of Markdown is not to *enforce* this, but rather to *enable* it. And I'll say it's a success: most Markdown documents are indeed publishable as-is. In some circumstances however, the author has to make some efforts or find some tool to keep things pretty (multi-paragraph list items and blockquotes comes to mind). If you don't intend to publish the Markdown version, there is not much point to this effort and, thankfully, you can just skip it.


> Regarding ease of editing :

> The difficult with inserting text into a column is a general problem with

> text editing tools and table formats in general.  It is not a specific

> problem with the proposed table syntax.


My point about editing is that it's much easier to edit the HTML table than your table syntax because there is no grid to maintain.


> Regarding cell alignment :

> In my original post I wrote this

>> The author has already provided the desired text alignment in the original

>> (mono spaced) markdown text.

>> 

>> It is therefore plausible for a parser to derive cell alignment by comparing

>> the amount of leading and trailing white space in each table cell of each row

>> and each column.

> 

> I am the first to concede that this would require near-perfect spacing in

> the document, and would be very hard to implement.  It is therefore unlikely

> that anyone would bother to implement it.


Alignment deduction would likely be error prone too.


> However, there's no reason not to include MMD-style cell alignment

> meta-characters in the specification as a more practical short-cut if that

> is what people want.


Indeed.


> Thanks again for your comments Michel -- I hope I was able to communicate my

> answers effectively and politely.


It's an interesting discussion.

-- 
Michel Fortin
michel.fortin at michelf.com
http://michelf.com/