mastodon - Link to source

For the next #Conversations_im release I’m refactoring how URIs are linked / made clickable. I’m adding a bunch of URI schemes like tel and mailto on top of the existing xmpp, http(s) and geo but removing support for "things that look like web URLs but aren’t actually URIs" (like 'example.com') to avoid some false positives.

Once the 2.18.0-beta comes out tomorrow or so let me know if you see things that isn’t matched and should be matched or vice versa.

A screenshot of a message bubble in #Conversations_im with the text: Hi. I'm refactoring how URIs are linked in Conversations. We now support more URI schemes like mailto:user@example.com and tel:+1-269-555-0107 and obviously maintain support for things like xmpp:conversations@conference.siacs.eu?join and https://example.com however we no longer link domains that aren't actual URIs like example.com to avoid some false positives. the mailto, the tel, the xmpp and the https links are underlined to signal that those are clickable links.
#conversations_im
in reply to Daniel Gultsch

mastodon - Link to source

mistersixt

I would like to tell you about a request that I have already received several times from friends and family members: you can search for terms in the chat history. The search itself works fine, but you can't jump from a result to the actual place in the chat history. But that would be important because the messages before and/or after the search term are often just as important. Currently, you have to remember the date, switch to the chat history and then scroll (possibly indefinitely).
Unknown parent

mastodon - Link to source

Guus der Kinderen

I wouldn't know how to do that either, to be honest. I'm guessing that this is a wheel already been invented by re-usable code, but maybe not for your platform. Perhaps consider dropping certain characters that are technically valid if they appear as the last character of the URI?
Unknown parent

mastodon - Link to source

allo

The computer scientist in me says you can't find matching parentheses with RegEx.

The programmer in me says look for a solution that is "good enough". An idea that would cover a lot of cases would be, for example, to assume that a URI can contain at most one pair of parentheses. The few counterexamples are rare, and they are then the rare false negatives one has to accept.

@rakoo @guusdk

@rakoo @Guus der Kinderen