New link syntax

[european bob]

On Wed, 2004-03-31 at 17:47, Lou Quillio wrote:
> No, I was thinking of an array in the the source code, to contain the 
> characters significant to MD.  On reflection this isn't so valuable:  
> the characters in use seem to be reliably language-neutral, and it may 
> not be wise to *promote* hacking the syntax.  Was just a thought.

Ah, gotcha, the word 'delimiter' should have really given it away ;) To
be honest, it's not such a terrible idea on the face of it, but probably
only on the basis that people would be able to test changes and use them
before proposing it for acceptance. For example, "this is a quote" is
quite natural to an Anglophile, whereas someone else might find «this is
a quote» more useful. Being able to hack the syntax is definitely a plus
in my book; anyone doing it will understand that what they're doing
(since they're more than likely going to be altering regexes; even if
they're configuration variables or something).

> > could/should a Markdown interpreter generate errors,
> > and how should they be passed to the user?
> 
> Opposed.  Shifts responsibility from the user to MD.  There'd be no end 
> to it.  What's an error?

An error is something a program emits when something happens that the
program thinks shouldn't. I don't think it shifts responsibility; it
merely says "oi! I think you've made a mistake here". For example, if I
defined two links within a document (e.g., [more examples]) I don't
really want to have to check the document myself. If I've written:

Go look at the [first test run] I did (there are also some
[more examples] of the test)

[more examples]: http://www.example.com/examples "My initial examples"

.. and then 1273 lines later I write:

So, the [final examples] show the result quite clearly, although there
are a few that [run contrary][more examples]

[more examples]: http://foo.example.com/examples "The final test run"

... preferably Markdown is going to tell me about that. I don't want my
initial examples link to be replaced by my final examples, or
vice-versa. Yes, the user is responsible for naming without collision,
but no, I don't think that means it's the user's fault when they make a
mistake.

The problem is basically because the user is making use of link
management afforded by Markdown. There are two ways of storing your
links: either interspersed in the document (easiest for people reading
the text) or all at the end (easiest for the person managing the
document, in terms of ensuring there is no collision). If they are
interspersed, best for the reader [in text], then the user has to
attempt to remember all the bits of text they have used, or some up with
some bizarre collision-resistant naming scheme. Probably okay when
you're writing a document, likely sucks when you edit it about three
months later. Humans are good at coming up with the same words time
after time.

Probably, I didn't want the word "error" first time around. "Warning" is
perhaps better. The whole point is that I don't want an edit->preview
cycle; I just want to edit the doc and be done. If there is a potential
problem with the doc, I want the parser to tell me as soon as possible.
I think enough of those problems could be detectable. I certainly don't
want to have to check the URLs are correct; that's tedious.

--bob.