[worklog] APZ bugs, Some webcompat bugs and HTTP Refresh

Monday, January 25, 2016. It's morning in Japan. The office room temperature is around 3°C (37.4F). I just put on the Aladdin. The sleep was short or more exactly interrupted a couple of times. Let's go through emails after two weeks away.

My tune for WebCompat for this first quarter 2016 is Jurassic 5 - Quality Control.

MFA (Multi Factor Authentication)

Multi Factor Authentication in the computing industry is sold as something easier and more secure. I still have to be convinced about the easier part.

WebCompat Bugs

• Bugzilla Activity
• Github Activity
• In Bug 1239922, Google Docs is sending HiRes icons only to WebKit devices because of a mediaqueries @media screen and (-webkit-min-device-pixel-ratio:2) {} in the CSS. One difference is that the images are set through content outside of the mediaquery on a pseudo-element, but inside directly on the element itself. It is currently a non-standard feature as explained by Daniel Holbert. We contacted them. A discussion has been started and the magical Daniel dived a bit more in the issue. Read it.
• Google Search seems to have ditched the opensearch link in the HTML head. As a result, users have more difficult to add the search for specific locales. Contacted.
• Google Custom Search is not sending the same search results links to Firefox Desktop and iOS9 on one side than Firefox Android. Firefox Android receives http://cse.google.com/url? instead of http://www.google.com/url?. This is unfortunate because http://cse.google.com/url? is a 404.
• DeKlab is sending some resources (a template for Dojo) with a bogus BOM. Firefox is stricter than Chrome on this and rejects the document, which in return breaks the application. Being stricter is cool and the site should be contacted, but at the same time, the pressure of it working in Chrome makes it harder to convince devs.
• When creating a set of images for srcset, make absolutely sure that your images are actually on the server.
• Some bugs are harder to test than others. redbox.com is not accessible from Japan. Maybe it's working from USA. If someone can test for us.
• When mixing em and px in a design you always risk that the fonts and rounding of values will not be exactly the same in all browsers.
• There is an interesting bug with Font being truncated on twitter differently on linux and other platforms and this depending on the fonts. Boris Zbarsky suspect that twitter plays somehow with the font. I started to dig to extract more information.
• Strange issue on a 360 video not playing in IceWeasel. This seems related specifically to the Debian version.
• Some of the bugs we had with Japanese Web sites are being fixed, silently as usual. Once we reached out, not always the person will tell us back, hey we deployed a fix.
• A developer is working on a Web site and he feels the need because of a performance, feature issue to create a script to block a specific user-agent string. Time passes. We analyze. Find the contact information given on the site. Try to contact and don't get any answers. We try to contact again this time on twitter. And the person replies that he is not working anymore there so can't fix the user agent sniffing. Apart of the frustration that it creates, you can see how user agent sniffing strategy is broken, it doesn't take evolution of implementations and human changes in consideration. User agent sniffing may work but it is a very high maintenance choice with consequences on a long term. It's not a zero cost.

APZ Bugs

There's a list of scrolling bugs which affects Firefox. The issue is documented on MDN.

These effects work well in browsers where the scrolling is done synchronously on the browser's main thread. However, most browsers now support some sort of asynchronous scrolling in order to provide a consistent 60 frames per second experience to the user. In the asynchronous scrolling model, the visual scroll position is updated in the compositor thread and is visible to the user before the scroll event is updated in the DOM and fired on the main thread. This means that the effects implemented will lag a little bit behind what the user sees the scroll position to be. This can cause the effect to be laggy, janky, or jittery — in short, something we want to avoid.

So if you know someone (or a friend of a friend) who is able to fix or put us in contact with one of these APZ bugs site owners, please tell us in the bugs comments or on IRC (#webcompat on irc.mozilla.org). I did a first pass at bugs triage for contacting the site.

• python docs website needs a patch
• MediaTek Web site needs a fix
• When working on this scrolling bug, I discovered completely unrelated to the issue that they were using the HTML name attribute for storing a JSON data structure with quotes escaped so that they would not be mangled with the HTML. Crazy!

Some of these bugs are not mature yet in terms of outreach status. With Addons/e10s and APZ, it might happen more often that the Web Compat team is contacted for reaching out the sites which have quirks, specific code hindering Firefox, etc. But to reach out a Web site requires first a couple of steps and understanding. I need to write something about this. Added to my TODO list.

Addons e10s

Difficult to make progress in some cases, because the developers have completely disappeared. I wonder in some cases if it would not just be better to complete remove it from the addons list. It would both give the chance to someone else to propose something if needed by users (nature doesn't like void) or/and the main developer to wake up and wonder why the addon is not downloadable anymore.

Webcompat Life

We had a https://wiki.mozilla.org/Compatibility/Mobile/2016-01-26. We discuss about APZ, WebCompat wiki, Firefox OS Tier3 and WebkitCSS bugs status.

Pile of mails

• Around 1000 emails after 2 weeks, this is not too bad. I dealt in priority with those where my email address is directly in the To: field (dynamic mailbox), then the ones where I'm in needinfo in Bugzilla, then finally the ones which are addressed to specific mailing-lists I have work to do. The rest, rapid scann of the topics (Good email topics are important. Remember!) and marking as read all the ones that I don't have any interests. I'm almost done in one day with all these emails.

HTTP Refresh

• Decided to finally open an issue on documenting HTTP Refresh. I had written about this HTTP header in the past. As usual, this is a big mess.

• explain what is necessary to consider a bug "ready for outreach".
• check the webkitcss bugs we had in Bugzilla and WebCompat.com
• check if the bugs still open about Firefox OS are happening on Firefox Android
• testing Google search properties and see if we can find a version which is working better on Firefox Android than the current default version sent by Google. Maybe testing with Chrome UA and Iphone UA.

Otsukare!