On 3/8/07, Steve Bennett stevagewp@gmail.com wrote:
Are you suggesting that you might end up with a chunk like:
<XML> ... <UNRECOGNISED> <CHUNK text="Some unrecognised code with {{funky!syntax|not..supported|by]]] wikiwyg" /> </UNRECOGNISED> ... </XML>
The concept sounds vaguely doable, but pretty much any comments not from the main mediawiki developers aren't even worth $0.02.
I think that's kind of excessive. :) "Main" MediaWiki developers are unlikely to be the ones to do this, and MediaWiki developers altogether maybe fifty-fifty at best, assuming it ever gets done. Remember that this is an open-source project.
Overall, I think this is probably a case of too much talk, not enough work. If this is going to get done, someone needs to get cracking. And I don't exclude myself here, although this isn't something I plan to try in the immediate future anyway.
On 3/8/07, Steve Bennett stevagewp@gmail.com wrote:
Isn't there a simpler way, something like:
{{subst:tilde}}{{subst:tilde}}{{subst:tilde}}{{subst:tilde}} ?
On 3/8/07, Alexander Ian Smith ais523@bham.ac.uk wrote:
{{subst:NonExistingPage}} doesn't transform on save.
Both of these are definitely bugs, and they should be fixed (although they aren't high-priority, obviously). Pre-save transforms should be done recursively with sanity checks and cleaning to preserve idempotence. For the most part, they *are* idempotent, and any cases where they are not are ipso facto bugs.
On 3/8/07, Timwi timwi@gmx.net wrote:
To be honest, I keep seeing talk about "defining the parser's behaviour", but I haven't seen any actual work on it at all. (I'm not saying there hasn't been any, only that I haven't seen it.)
There's been some work that I've seen, mostly abortive. I think your flexbisonparse probably constitutes most of the applied work people have done on it, though. I think other things I've seen have been people trying to work out grammars, but not actually getting them to work with bison or whatever.
On 3/8/07, Timwi timwi@gmx.net wrote:
Sheldon Rampton wrote:
(1) The XML would be close to the HTML version that most people see when they view articles in their web browser, so less parsing would need to be done when presenting articles for viewing.
This conclusion is incorrect. It doesn't matter how "close" they are to each other, you would still need to parse it.
But you would need to parse it "less", i.e., you could parse it more simply and quickly, with fewer rules and no informal rules. And you'd also be parsing it in a more standard way, using XML functions, so you can get C-like (at least C++-like) speed even in, say, JavaScript, which I believe now comes standard with lots of XML functions at least in most implementations. So something standard like XML has advantages. You could even use XSL for the transform rules, to avoid having to write your own transformations.