Henric Persson

Here's an idea for how we could aggregate results: Instead of serializing the result json into the uploaded html document, we provide a separate json file with the results. We...

True. I don't have too much experience setting up systems like these. I was mostly thinking that the upload from workers would be simpler than having to report back to...

This makes sense to me. I'm trying to think about ways this might be specific to s3 in any way, but I don't see any. Any happo "reporter" could be...

This makes sense. I think we could allow `upload` to take an `id` of some sort (similar to what's discussed here: https://github.com/Galooshi/happo/issues/73#issuecomment-268190554) and overwrite the old results page if one...

Are you sure we shouldn't use the project you've been working on? We should pair this with setting up happo diffs in Travis CI for our components.

One idea that would get this going sooner without being blocked on transferring to a node project is to expose a few scripts that you can use. ``` happo travis-ci...

Just a thought: If we decide to implement this as npm modules, and allow users to plug them in using `.importjs.js`, we should make sure that plugins are explicitly imported...

**Running in Firefox** happens in our internal Travis setup, where we indeed do exactly what you suggest: ``` yaml before_install: - "export DISPLAY=:99.0" - "sh -e /etc/init.d/xvfb start" ``` We...

A first version could just use what we have with S3 today. Post a link to the results. We can treat the inline image posting as an enhancement.

It would be awesome if we could get a diffux run for diffux-ci components. I think if we work on https://github.com/diffux/diffux_ci/issues/61 we might actually have something to diff.