hypothesis / vision

Envisioning the future of the Hypothesis.
https://github.com/hypothesis/vision/issues/
40 stars 8 forks source link

Support @Mention #147

Open JakeHartnell opened 9 years ago

JakeHartnell commented 9 years ago

As a user I would like to receive a notification when someone mentions my username.


@dwhly originally opened https://github.com/hypothesis/h/issues/580 with the following description:

Invoking another user in an annotation via the @ syntax, i.e. "@username" would leverage a commonly understood mechanism in other systems for inviting that user to participate (ala twitter, g+, et al). We should probably implement.

In that issue @csillag made a proposal which was +1'd by @tilgovi:

How about we interpret @user as user@SERVER[:HOST], but still allow and recognize mentions in the format of @user@other-domain.com, too? That way, we would not impose any constraint, but still have basic convenience for 99% of the cases. (Although the parsing might be tricky, because of the double role of the at sign...)

JakeHartnell commented 9 years ago

Related to https://github.com/hypothesis/h/issues/763.

csillag commented 9 years ago

I think this is exactly the same as https://github.com/hypothesis/h/issues/580

JakeHartnell commented 9 years ago

I wanted something that was cleaner and in vision, #580 was closed and #763 should be an epic. I am happy to close this if need be, or alternatively can we close #763 and move it to vision as an epic? Thoughts?

csillag commented 9 years ago

I think closing https://github.com/hypothesis/h/issues/580 was a mistake, because it's not really a duplicate of https://github.com/hypothesis/h/issues/763. I propose copying (much of the) content from https://github.com/hypothesis/h/issues/580 to here, maybe with some simplifications.

dwhly commented 9 years ago

I'm happy to leave this open here now in vision.

BigBlueHat commented 9 years ago

Dropbox just announced sidebar comments + @ mentions.

Type @ and you can auto-complete from full names and email addresses (which is the "vector" used for sharing with Dropbox): Dropbox @ Mentions

Once selected the full name of the user is put in place of the @whatever text (similar to the Facebook mentions experience): @ Mention suggestion selected

Nice thing here is that user identifiers are technically distributed. The auto-complete is from within the scope of shared people. However, you can also mention email addresses: @new@example.com and on pressing the spacebar, the email address (rather than a name) will appear in the field: @ mention an email address Note the text under the field has changed to explain that "1 person" will be notified.

So, the take away for us:

Just :bulb:'s. :tophat:

dwhly commented 9 years ago

invite any email address and they should get a "You've been invited to Hypothes.is" email

@judell and I discussed exactly this implementation a few weeks ago. My thought is that until the person accepts the invite-by-mention and gets a username, their email should be obscured or mangled in some way. Alternatively, you could ask for a handle (of some sort) in addition to their email when mentioning, and the handle could be shown until they claim their own.

dwhly commented 8 years ago

@tilgovi:

I do think share via email is a great feature. I imagine there's been some change since about five years ago, when I last looked at these sorts of metrics, but it used to be the case that email way surpassed Twitter, Facebook, et al for "social" sharing. And it's nice to not have the "NASCAR effect" of something that shows a dozen different share icons. Obfuscation of emails would be a good idea, but there're more reasons to be concerned by such a feature that should be taken into account.

There's the recent discussions about Genius and the potential for annotation to be abusive or harassing. Letting users send emails to arbitrary addresses opens up a vector for abuse because there's no opt in to such a thing by design. Possibly there's an opt out, if the user signs up and then disables notifications. There's also the problem of sending email to arbitrary addresses that puts the reputation of Hypothesis on the line to prevent it being used by spammers.

So, perhaps with...

.. could be a viable feature.

On the other hand, there's always mailto: which, beyond lifting these risks off Hypothesis, probably also results in higher open rates because the email would come from the person sharing rather than from Hypothesis. It's also quick and easy to implement. The only downside is that it requires space in the interface.