The comment reply system is, at first glance, a fairly elegant and simple system that solves a fundamental problem with comments. However, there are many pitfalls and edge cases that even experienced users may not notice.
Take my own Allow for Comment Reply to Usernames With Less Than Three Characters in First Word for instance - when I first posted that question, nobody understood that particular aspect of the system, even though it was clearly documented in the faq.
I propose we introduce a simple and unobtrusive system of warnings that appear as users type. Right now, there is no indication of success or failure to the person typing the comment. This means that if they get it wrong, there's no way the error can be addressed unless someone goes up to him and say 'you're doing it wrong'. This is both a waste of time, and adds unnecessary noise to the already noisy comments.
I suggest that we start off with a simple message that simply informs the user of the system. This can be turned off for users with higher reputation:
Simple client side JavaScript can be used to determine whether the @username
the user has typed is valid or not, and a simple simple warning can be shown if he makes a mistake:
Suggestions such as the one shown above can also be determined intelligently. Another typical issue found in comments is that the commenter will often try to address two different users in a single comment. The second @username
will be ignored, but the commenter may not realise that is the case. This can also solve the problem:
In addition, other forms of feedback can also be used. For instance, @username
s in existing comments can be highlighted, if only ever so slightly, and a tooltip be used to highlight what it's for:
This will aid greatly in the discoverability of this feature, and since only successful use of the system will be picked up here, it should be plain to see if he has done anything wrong and need to edit. And always, there's the option of adding a 'help' link to the commenting interface, if only for new users:
The key too all of the above suggestions is that there must be feedback to the user on whether or not his action was successful, or in this case, unsuccessful, and it is this aspect in which I find the current system lacking.
@Y
it brings up all users who've contributed to the thread that start withY
.[feature-req]
- it's 3am now and I need to (yawn) sleep. Oh, and (shameless plug) rchern's SE mod script already does this