Provide links to relevant MDN articles in comments

[openjck]

No description provided.

[openjck]

I would be interested in observing users before committing to this feature. I expect the current warning messages will be sufficient in almost all cases. After all, we're only flagging compatibility issues. MDN is great for correcting usage errors (syntax, acceptable values, etc.) but we're not detecting that sort of thing. I'm not sure what our documentation would add.

We can reopen this issue if this need becomes clear in real-world usage.

[groovecoder]

Somewhat related to #48 ...

If YouShouldUse is built on the Fogg Behavior and Hooked models with an emphasis on triggers/notification that flow into action, reward, and investment, then we should watch how users respond to a full cylce of that.

Without this or #48, the commit comment does not lead towards any particular action. I'd be happy to watch what users do with the commit comment, but the reality is that we simply can't, unless we're physically sitting behind them?

If we provide links back to MDN, complete with acquisition parameters, we can see how users use (or don't use) the actions we build into our triggers.

[openjck]

I agree that we need to see users acting on compatibility comments. Discord can't achieve its goals if people ignore it.

I disagree that compatibility comments don't lead to any particular actions. If something as simple as a phone buzzing can inspire someone to check their messages, even without a specific call-to-action to that effect, it seems reasonable that documentation about browser support would be enough to inspire someone to change their code.

After all, compatibility comments are just line comments. I almost never provide links in my line comments, but people almost always address them anyway.

Ultimately, we want users to fix incompatible CSS. Can we measure that directly?

[groovecoder]

Back to this since #15 is closed.

We should measure CSS fixed directly then. That's how we'll watch users to determine if we're changing behavior or not.

In theory, we should be able to do this with the tool itself? Though we would have to start recording the # of issues found in each PR, which means we have to start storing results somewhere.

[openjck]

Definitely, it sounds like a database is in our future.

I think measuring corrections would be great! 😄

[groovecoder]

Based on my experience with the discord comment on MDN Tabzilla PR we should re-consider this. A link to MDN CSS media queries Browser compatibility could have done the same thing that @alexgibson did - i.e., explain that Safari would use the device-pixel-ratio.

[openjck]

Yeah, we should keep considering it. We definitely want to address needs that arise in the real world.

I would love to help us study users over time to understand how common the need is. We should be able to get an even clearer picture once we have more users and users with different backgrounds.

Somewhere there's a sweet spot between scope and maintainability. I would love to help us puzzle out its location so that we can have the best of both worlds.