How about this:
[http://www...] renders (a href="http://www...%22)%5B1%5D(/a) with autonumbering, [http://www...%7Ctext] renders (a href="http://www...%22)text(/a) (the pipe can optionally be a space), and [http://www...%7C] (note trailing pipe) expands to [http://www...%7Cwww...] ?
Why not just render http://www... as www... ? I've done that for my own wikis and it looks very nice, also for mailto:'s etc.
If we just keep adding and adding more wiki syntax on top of the old, we'll end up with an incomprehensible mess and code that's impossible to debug. We're going to have to simplify the markup before we extend it, and interpreting URLs outside of brackets is the most egregious problem in my opinion--it's an exception to the other forms of markup, it causes the punctuation problem, and it causes real coding headaches (like the bug of URLs with contained URLs, for example).
For a small, limited-purpose wiki, a limited feature set is fine and it doesn't matter if the markup language--which doesn't need to have a lot of features--is a bit awkward. But we've outgrown that. We're going to need a real markup syntax, consistent, predictable, documented, bug free, and with all the features needed for a complete general-purpose encycopedia.
Far better we should go through the pain of adjustment that will entail now, than later after we've got even more people to confuse.
"We're going to need a real markup syntax, consistent, predictable, documented, bug free, and with all the features needed..."
But we MUST NOT violate the sacred principle of Not Invented Here, which strictly forbids the use of the markup language that has real markup syntax, consistent, predictable, documented, bug free, and with all the features needed ... HTML.
</sarcasm>
-----Original Message----- From: wikipedia-l-admin@nupedia.com [mailto:wikipedia-l-admin@nupedia.com]On Behalf Of lcrocker@nupedia.com Sent: Friday, August 09, 2002 15:01 To: wikipedia-l@nupedia.com; The Cunctator Subject: Re: Re: [Wikipedia-l] Transitioning toward a clean wiki syntax.
How about this:
[http://www...] renders (a href="http://www...%22)%5B1%5D(/a) with autonumbering, [http://www...%7Ctext] renders (a href="http://www...%22)text(/a) (the pipe can optionally be a space), and [http://www...%7C] (note trailing pipe) expands to [http://www...%7Cwww...] ?
Why not just render http://www... as www... ? I've done that for my own wikis and it looks very nice, also for mailto:'s etc.
If we just keep adding and adding more wiki syntax on top of the old, we'll end up with an incomprehensible mess and code that's impossible to debug. We're going to have to simplify the markup before we extend it, and interpreting URLs outside of brackets is the most egregious problem in my opinion--it's an exception to the other forms of markup, it causes the punctuation problem, and it causes real coding headaches (like the bug of URLs with contained URLs, for example).
For a small, limited-purpose wiki, a limited feature set is fine and it doesn't matter if the markup language--which doesn't need to have a lot of features--is a bit awkward. But we've outgrown that. We're going to need a real markup syntax, consistent, predictable, documented, bug free, and with all the features needed for a complete general-purpose encycopedia.
Far better we should go through the pain of adjustment that will entail now, than later after we've got even more people to confuse.
[Wikipedia-l] To manage your subscription to this list, please go here: http://www.nupedia.com/mailman/listinfo/wikipedia-l
On Fri, Aug 09, 2002 at 02:58:18PM -0700, Sean Barrett wrote:
"We're going to need a real markup syntax, consistent, predictable, documented, bug free, and with all the features needed..."
But we MUST NOT violate the sacred principle of Not Invented Here, which strictly forbids the use of the markup language that has real markup syntax, consistent, predictable, documented, bug free, and with all the features needed ... HTML.
</sarcasm>
Apart from the fact that you forgot an opening tag you are right of course. If consistency, et cetera, was all we wanted then HTML (I'd actually prefer XML) would be an excellent solution. There are even more advantages: - There are good free XML parsers in PHP out there, so writing a parser would be relatively trivial and it would be easy to get total control over what XML is allowed. - Documenting the mark-up could simply be done by defining a DTD. - Combined with CSS and/or XSL we would have an excellent way to separate content and presentation. - Our internal format would at the same time be the best possible export format. - Any extension we will think of in the future can be easily integrated then. (That's what the X in XML promises, anyway. :-))
On Sat, 10 Aug 2002, Jan Hidders wrote:
From a technical point of view XML is clearly superior. So why don't
we use it? Because we want a mark-up that is easy to read, easy to understand, easy to type and non-intimidating for people who know next to nothing about HTML.
I think a good way of viewing the problem is to try not to think of all our wiki kodes as "markup", but rather as features of our simple editor. When someone makes a GUI frontend to Wikipedia, those features will become more sophisticated (wysiwyg).
What we store in the database is an entirely different question, but right now, since we only have one frontend, storing the plain editor text there is easier. One day, it will probably be XML.
-- Daniel
lcrocker@nupedia.com wrote:
If we just keep adding and adding more wiki syntax on top of the old, we'll end up with an incomprehensible mess and code that's impossible to debug. We're going to have to simplify the markup before we extend it, and interpreting URLs outside
What if "http:" was a namespace like "user:", you would use [[namespace:<data>]] for all sorts of links and that would be it. No more bare URLs and no [URL] with single brackets. That would be a reduction in complexity, and give you a modular design, with a new module for each namespace that would interpret <data> following. You could have [[tex:<data>]] for TeX math formulas and [[ISBN:<data>]] for ISBN numbers. The only rule for <data> would be it couldn't contain a double closing bracket.
Existing links and patterns that fall outside of this would have to be converted, but data has been converted before.
I haven't tried this, and don't know if it would work out.
wikipedia-l@lists.wikimedia.org