Web-bugs: www.evernote.com - Firefox not supported

[MichaelKohler]

URL: https://www.evernote.com/client/web

Browser / Version: Firefox 63.0
Operating System: Mac OS X 10.13
Tested Another Browser: Yes

Problem type: Something else
Description: "Sorry, we don't support your browser at this moment. To use the new Evernote Web experience, please use the latest version of Google Chrome or Safari. For a more integrated experience, you can also download Evernote." ... The same happens in Release Firefox as well, so it's not just UA sniffing gone wrong.
Steps to Reproduce:

_From webcompat.com with ❤️_

[freaktechnik]

Can confirm. Works without issue when spoofing the UA in nightly with the responsive design mode to Safari on iOS.

[augbog]

Hey just to chime in here, we haven't gotten full sign off on specific browsers to release the new Evernote web client version so we currently just do basic user agent checks. If you want to spoof the user agent you can, but we can't guarantee everything is working as expected.

We are working to roll out support for other browsers in the near future and by no means was this meant to be a long-term thing. Sorry about that 🙏

[softvision-sergiulogigan]

Thank you all!
@adamopenweb I think it's safe to move this one to sitewait, right?

[julen]

Can confirm. Works without issue when spoofing the UA in nightly with the responsive design mode to Safari on iOS.

Expect different sorts of issues in the editor; the rest should be fine. Source: I've been working on Firefox support myself.

[adamopenweb]

Thank you all!

+1

The site appears to be working in Firefox Nightly 63 on MacOS. If someone can confirm, let's close this report.

[MichaelKohler]

The site appears to be working in Firefox Nightly 63 on MacOS. If someone can confirm, let's close this report.

Negative:

I think @julen's comment was just referring to spoofing the UA.

[softvision-sergiulogigan]

We are working to roll out support for other browsers in the near future and by no means was this meant to be a long-term thing. Sorry about that

Let's wait for this...

[adamopenweb]

Apologies if I've confused things here. But it seems the site is working for me in Nightly and release.

Mozilla/5.0 (Macintosh; Intel Mac OS X 10.13; rv:63.0) Gecko/20100101 Firefox/63.0

[adamopenweb]

Edit: My mistake this is the new version we are looking at. I can reproduce. Sorry! 😊

[adamopenweb]

Hey @augbog do you have an update on support for Firefox that you can share? 😄

[adamopenweb]

Wondering if @augbog or @julen has any additional info they could share on FF support? Thanks!

[augbog]

Hey really sorry thanks for holding me accountable! Let me reach out to a PM to get some more insight. I will try to have a response by tomorrow.

Once again super sorry about that.

[adamopenweb]

Thanks for all your help 😃

[augbog]

Hey so I synced up with a PM yesterday. The target date right now is June for Firefox readiness but don't want to make any promises. The editor we are using had some bugs and they were severe enough that we did not want to open the new client up for Firefox yet. I will try to post more updates if possible

[emilio]

Fwiw (I've done some work here and there re Firefox editing), if there are severe differences between browsers that make supporting Firefox harder, please by all means do file Firefox bugs, or cc me on the pre-existing ones. Knowing about compat issues is the only way we can fix them :)

[julen]

Hey @emilio, thanks for being open and supportive. The problem here mostly boils down to contenteditable and its lack of spec implemented across browsers, plus some other internal details about the way the current EN editor is implemented.

[emilio]

Sure, I'm aware, contenteditable is quite a mess.

I've been doing a fair amount of contenteditable-related interop work (see https://bugzilla.mozilla.org/show_bug.cgi?id=1506547 and co) and spec stuff (https://github.com/w3c/csswg-drafts/issues/3344).

I'm not fully devoted to that work, but I'm happy to help if possible. I'd still encourage to file issues against bugzilla or csswg-drafts (maybe? not sure who should really define contenteditable, css-ui defines a bunch of properties that interact with it), and cc me so I either fix stuff on Firefox or discuss with other vendors about what's the desired behavior and get it into a spec.

[advicepyro]

Just wanted to chime in that the only bug I see so far in Firefox is an extra space being inserted at the beginning when I open a new code block. Otherwise, the new beta seems to work okay with the UA string changed.

edit: I stand corrected, there are more bugs with regards to newlines; sometimes I get doubled newlines. The cursor is also misaligned when I add a new to do box.

[augbog]

Hey we're prioritizing some FIrefox bugs this sprint and hoping we can release support within the coming week(s)! @ilias-t has been doing a lot of work for this but hoping to provide more updates! :)

[ilias-t]

Thanks @augbog, to give everyone a bit more context, we've rebuilt our editor which has now made supporting Firefox possible. We're now doing a final pass at outstanding UI issues before we begin to officially support it 🦊

[adamopenweb]

This is great news @ilias-t! Thank you! 🎉

[cipriansv]

After retesting the issue I confirm that the issue has been fixed.

Tested with:
Browser / Version: Firefox Nightly 69.0a1 (2019-06-26)
Operating System: macOS Mojave 10.14.5

[julen]

@cipriansv that's the old web client, which already worked in Firefox. This issue is about the new client which is not Firefox-compatible yet — this is in the works, though, and might be out sooner rather than later. So please reopen the ticket, otherwise we'd be spreading misleading information. Thanks! :)

[ilias-t]

Ok it's been done! We should now be able to close this issue 🎉

A rebuilt version of the note editor is now the default editor for Firefox users. Again the issue with originally supporting Firefox was the fact that our old editor was using the the view as a source of truth; however, browsers represent that view state in different ways (Firefox has some idiosyncrasies how it handles content-editable input, which proved challenging to work around). The new editor does not have the same issue and can therefore support Firefox/all major browsers. This new editor has proven itself quite stable, so we are now comfortable promoting it a larger audience (all the Firefox users).

Sorry for this issue, and thanks for bearing with us while we addressed it. Enjoy!

[emilio]

Ok it's been done! We should now be able to close this issue

Awesome, thanks!

Firefox has some idiosyncrasies how it handles content-editable input, which proved challenging to work around

FWIW, are these documented somewhere? If so, could you please file bugs about those, and cc me on them? (emilio [at] crisal.io)

There's no way to get those fixed if we don't know about them :)

[julen]

@emilio I spent some time in the past trying to address some of the contenteditable issues on our editor's side, and came across a few specific issues. I'm not fresh on the details, but here are some of the bugs that I could find from my BZ account (CC'd you on these):

• https://bugzilla.mozilla.org/show_bug.cgi?id=1289673
• https://bugzilla.mozilla.org/show_bug.cgi?id=1239411
• https://bugzilla.mozilla.org/show_bug.cgi?id=1444323

[emilio]

Awesome, thanks @julen

[emilio]

All of those are about table + contenteditable, which is one of the funniest parts of the editor in Firefox. There is a bunch of code in Firefox that assumes selections have one single range.

Regarding the first, there's ongoing work in figuring out what the right model for selection crossing shadow boundaries is. There is a chance that we go for multi-range selection, in which case all those places should be fixed (and it'd fix multiple issues with tables too, I suspect).

The second looks like an issue with our selection movement stuff which is probably tricky and I won't have the time to look at soon, but I'm happy to mentor and point to code to people that want to give a shot at it.

The third is probably trivial to fix, I've pinged one of the editor folks to see what the right condition is here.