Recognize odd characters at end of URLs

Ticket

+commit
 

It makes sense that we should not process any of these at the end of a link that might be used at the end of a clause or sentence:

/* ! . ) ? , : ; */

However, other trailing characters should generally not be cut off.

 

 

 

Part of the problem was recognizing word endings right next to html tags (like

).

 

 

I fixed it for most of the common cases. However, if there is an   immediately after the link, there may still be problems. This can be worked around easily, so I suspect further improvements will need to be put off until after 1.0.

 

--Ethan


: still unrecognized at end, as are ! and ? and .

[ and ] now completely prevents URL recognition no matter where they occur in it, as do % ^ | { } " ` (note last is backtick, not single quote)

see http://sandwagn.wagn.org/wagn/Recognize_odd_characters_at_end_of_URLs

 

Not sure i agree with not recognizing these ! . ) ? , : ; at end of URLs. If a URL could end with them, then i'd lean toward recognizing them as part of the URL.

 

Anyway, if you still accept the current functionality for Wagn 1.0 just say so, and i'll move all this to a new ticket.

--John Abbe.....Tue Feb 10 16:12:13 -0800 2009


URL auto-recognition is just for convenience, and should work right in the common cases. No link at all is better than a busted one.

A url that ends in ? or . is a weird imaginary case that rarely happens. A sentence or clause ending in a url followed by punction is very common. "Have a look at my pictures at wagn.org." Your solution would break the common case by sticking unwanted characters on the end of the url. I don't really want to ticket that idea.

Given that it's harder to know what's supposed to happen when you get these other characters in the middle of a url, it may be best not to process them. It's better than what currently happens: TestOne. (I copied your examples over. Let's delete that soon.) So I'd say it's an improvement in that case, too.

Most of those characters will never appear in escaped urls. The one that does concern me is the %, which is extremely common. After that come [ and ], which are common in ruby urls. Perhaps we should ticket that one.

But autodetection, still, is a bonus. It's more vital that things work correctly in the explicit case. I suspect in most of these cases they do, but I worry about the ]. Here's a test:

[[http://wagn.org/new?card[type]=Image|new image]]

  --Ethan McCutchen.....Wed Feb 11 11:11:10 -0800 2009


wow. that wack.

OK, so my take is that there is definitely no more time for this ticket prior to 1.0. I personally think the changes made are improvements and should be kept. If you disagree, we can discuss rolling them back. Otherwise let's get these issues moved to the background.

  --Ethan McCutchen.....Wed Feb 11 11:17:18 -0800 2009


Closing. New ticket is improve URL recognition.

  --John Abbe.....Wed Feb 18 12:45:48 -0800 2009