Page MenuHomePhabricator

Human-readable URLs: Improve unique identifier for each feedback post on Special:ArticleFeedbackv5
Open, Needs TriagePublic

Description

Currently, it's a randomly generated random string of letters A-Z + a-z and 0-9, set to 30 characters total which is assigned to each feedback post. They don't make sense and are thus fail for human-readability. This is an example of what it looks like:

http://en.brickimedia.org/wiki/Special:ArticleFeedbackv5/Darth_Vader/055ba317c77597e9ad01f902f6477168

(Using most recent feedback posted by a registered user as of this time for the example, hope that's okay with you @lcf119 :P)

Solution idea 1

My proposal is to write it in a way where instead of a randomly generated string (while still providing the ability for unique identifiers to not run out on a high-traffic wiki with a large amount of articles and a large community) is this:

Special:ArticleFeedbackv5/{article}/{feedback author}/{#}

The concept is that it grabs the Feedback poster's username into that URL, and depending on how many feedback posts they've made, it'll increment that number (the #) by 1.

For example, we could convert the above URL to use the concept provided above. If this was Edward Nigma's first time posting feedback on Darth Vader, it would look like:

Special:ArticleFeedbackv5/Darth Vader/Edward Nigma/1

Solution idea 2

Here I'm listing another thought I had. This second one was the same as the first solution that I've provided but instead of that, it would be the date that that feedback was posted, in a date format which most people recognize (DDMMYYYY) but that would prevent the ability of two unique identifiers if say, <username> posted feedback on Darth Vader twice on the same date. We could possibly expand it to hhmmss (hours, minutes, second) which results in a syntax that looks like this:

Special:ArticleFeedbackv5/Darth Vader/Edward Nigma/DDMMYYYYhhmmss

This isn't as bad compared to the current URL architecture, since it's a username and a datetime format which is 14 characters long, compared to 30 characters (less than half) but the problem with is that it wouldn't be immediately recognizable just by looking, you would just see 14 numbers and it probably wouldn't be recognizable until it's directly explained which isn't very good. It's basically like the quote "if you have to explain the joke, it's not funny." but for user interfaces and design:

"A user interface is like a joke. If you have to explain it, it's not that good."

Replies with feedback and comments to these ideas are welcome!

Event Timeline

SamanthaNguyen renamed this task from Improve unique identifier for each feedback post on Special:ArticleFeedbackv5 to Human-readable URLs: Improve unique identifier for each feedback post on Special:ArticleFeedbackv5 .Sep 24 2016, 12:14 AM
SamanthaNguyen created this task.
SamanthaNguyen moved this task from Backlog to Feature Requests on the ArticleFeedbackv5 board.
SamanthaNguyen moved this task from Backlog to ArticleFeedbackv5 on the Brickimedia board.

Currently, it's a randomly generated random string of letters A-Z + a-z and 0-9, set to 30 characters total which is assigned to each feedback post. They don't make sense and are thus fail for human-readability. This is an example of what it looks like:
http://en.brickimedia.org/wiki/Special:ArticleFeedbackv5/Darth_Vader/055ba317c77597e9ad01f902f6477168

That is probably a hash of information compared to random (But I haven't looked it up).

Personally, I suggest looking at giving each submission a unique ID number, then the URL could for example just be Special:ArticleFeedbackv5/#####.

  • Solution one will require a count lookup every-time someone goes to submit feedback
  • Solution one and two will produce extra work when a user is renamed

I'm not sure human readable urls are all that needed in AFTv5, How often would people really be typing them in compared to accessing the feedback via the appropriate special page(/s)?

Thanks for pointing those out, makes sense. I like your idea better :)

Also, probably not often, but shortening (hopefully as much as possible) would still be a considerable improvement compared to the current nonsensical 30 char string (just from the looks of it). It /might/ have some meaning, but the meaning isn't very recognizable at a first glance.