I also wanted to take this opportunity to make some more comparisons about Google Tasks vs. Asana that weren't really appropriate for the essay:
- It's annoying that Asana does not support real hierarchy. Asana has responded to this on Quora, but their response is really about making Asana's life easier, not yours. What I think is really interesting is that if they ever decide to reverse their decision, it will really screw with their keyboard shortcuts since they currently heavily leverage Tab as a modifier, but that may not work so well when users expect Tab to indent (and un-indent).
- It's pretty subtle, but when you check something off of your list in Google Tasks, there is an animated strikethrough. I did this by creating a timing function that redraws the task with the </s> in an increasing position to achieve the effect. This turned out to be tricky to get right because you want the user to be able to see it (independent of task text length) without it feeling slow for long tasks. Since so many people seem to delight in crossing things off their list, this always seemed like a nice touch in Google Tasks. I don't find crossing things off my list in Asana as satisfying.
- Maybe I haven't given it enough time yet, but I have a lot of trouble with Asana's equivalent of the details pane. The thing seems to slide over when I don't want it to, or I just want to navigate through my tasks and have the details pane update and it starts bouncing on me...I don't know. Am I alone? In Tasks, it's shift-enter to get in and shift-enter to get out. It's not as sexy since there's no animation, but it's fast, it works, and it doesn't get in my way.
Finally, some friends thought I should include a section in the essay about "what browser vendors can do to help" since clearly the browser does not have the right APIs to make things like Google Tasks easy to implement. As you may imagine, I was getting pretty tired of writing the essay, so I didn't include it, but this is still an important topic. At a minimum, an API to return the current (x, y) position of the cursor as well as one to place it at an (x, y) position would help, though even that is ambiguous since a cursor is generally a line, not a point. It would be interesting to see how other UI toolkits handle this.
Oh, and I should mention that if I were to reimplement Google Tasks again today, I would not implement it the way that I did, if I had enough resources. If you look at Google Docs, you will see that your cursor is not a real cursor, but a blinking <div>. That means that Docs is taking responsibility for turning all of your mouse and keyboard input into formatted text. It eliminates all of the funny stuff with
contentEditable by basically rebuilding...well, everything related to text editing in the browser. If I had libraries like that to work with (that already have support for pushing changes down to multiple editors in real time), then I would leverage those instead. Of course, if you don't have code like that available, then that is a pretty serious investment to make. Engineering is all about picking the right tradeoffs!
Want to learn more the technologies used to build Google Tasks? Pick up a copy of my new book, Closure: The Definitive Guide (O'Reilly).