Hacker News Reader

Comment by endymi0n | Hacker News Reader

While you're completely on the mark with this, I noticed that Google pulls the new Microsoft now that it has almost a quasi monopoly on the desktop: Most of the new, complex UIs by Google services (read: GCP Cloud Console, all of Google Marketing Platform) are slowing my Firefox to a grinding halt, often times even crashing the tab.

No problems in Chrome whatsoever. It's the only thing I'm using Chrome for these days and I can't help shake the same ugly feeling than when I was forced using Internet Explorer for government websites purely because of their ActiveX plugins back in the days.

View on HN

I've ran into various comments around the internet/reddit that Google sites like Youtube are slower on Firefox, and that changing the user agent to Chrome fixes it.

No problems here: Waterfox plus usual privacy oriented addons then Hooktube replacing Youtube. Might seem excessive but works like charm.

https://www.waterfoxproject.org/en-US/

https://hooktube.com/

https://addons.mozilla.org/en-US/firefox/addon/ublock-origin...

https://www.eff.org/https-everywhere

https://www.eff.org/privacybadger

https://addons.mozilla.org/en-US/firefox/addon/umatrix/

https://addons.mozilla.org/en-US/firefox/addon/hooktube-redi...

Just a heads up—HookTube's homepage (at least for the moment?) is very NSFW.

Warning: Hooktube link is NSFW. 'Sensual massage' thumbnail on the frontpage...

Just checked, you're absolutely right! I always use the redirector addon rather than going to their homepage though and it has been SFW so far.

To all, I can't edit anymore my post to delete it, so just forget about the hooktube address; as others reported it's definitely NSFW while the redirector Firefox addon is safe. I apologize for not checking it before.

That seems hard to believe as it would imply that they have code specifically to cripple the Firefox experience. It's probably at worst that they intentionally don't optimize their sites for Firefox and not actively making it worse.

> That seems hard to believe as it would imply that they have code specifically to cripple the Firefox experience.

I think it's more the case of enhancing the Chrome experience and not doing anything to deliberately impact firefox.

That is the same as what I suggested. Which would mean that simply changing a user-agent can't make anything run better on firefox.

Google does a lot of UA sniffing, and serves different sites to different browsers.

The sites served to non-Chrome receive less QA resources, as far as I can tell, and are often buggy in various ways. Simply spoofing the Chrome UA in Firefox on Android can often get you a site that works better (in Firefox) than the one served to Firefox by default.

Sometimes this happens on the desktop as well, they optimize for the latest Chrome and the "latest" Firefox ESR - so here we are, Chrome is the reincarnation of IE6'6'6 the definition of evil in the software world.

I've noticed this also. Gmail or anything by Google within Firefox is painful. Very very painful.

Just a heads up, but we are actively working on improving the situation here, and should have releases coming out that steadily and progressively improves the experience on this front. It is an area of major focus after the initial Quantum Flow effort, and we have technical fixes in many places in the pipeline to address this.

(I work on the Javascript engine in Firefox, and improving our story here is one of my personal top priorities, as well as an organizational priority).

Are you saying that performance issues on Google properties are a bug in Firefox? The other posters were implying the issue was Google doing things to intentionally slow down performance in Firefox.

I'm not about to speak to the intent of programmers I haven't interacted with heavily. In these sorts of charged conversations there is often an impulse to make issues about "this" or "that" exclusively. I find it useful to avoid that impulse entirely and instead focus on what I can do to make things better.

Well then let me restate my question. Are there regressions in Firefox that hinder performance on Google properties?

Not stating Google is intentionally slowing down things - but I've noticed that Google products and only Google products are slow to render, slow to click into the search field of Google Maps, and when I press 'compose' within Gmail on a macbook pro 2017 I can wait easilly 10 seconds before the email bit pops up. Within Chrome however, quick and slick.

Google has a history of disabling or degrading features (image search, maps, youtube) in browsers that aren't Chrome under the "we haven't fully tested them" excuse. In almost all cases changing your user agent to Chrome would result in a perfectly functional site. Perhaps there is 1 feature that didn't work but it was usually obscure and it feels like Google was just throwing the baby out with the bathwater to spite anyone not using Chrome. Up until recently it felt like Google made a Chrome version of their product and a lowest common denominator version and anyone that isn't using Chrome got the later. Now it seems like they let some browsers use the Chrome version but there are unusual performance regressions compared to Chrome. It's very reminiscent of the first browser wars where sites were IE or Netscape only.

The whole point of standards push over the years was so that developers could just code to the standard and the browsers would either catch up or suffer the consequences. It seems like these days Google is dipping into Microsoft's old playbook for every trick it can find to get people to switch to Chrome.

If you want a site the is just horrible in Firefox Mobile, try: http://www.bom.gov.au/australia/meteye/ This is a Australian Government site, so there is no Google shenanigans going on here.

You are meant to zoom in on a locality on the interactive map, and then pan around by dragging. Panning works fine in Chrome. The panning is so slow in Firefox Mobile it's unusable . (It's fine on Firefox desktop, but I'm guessing the shear horsepower available on a desktop CPU hides the problem).

It's a bit sad, because it's a page I use a lot. Whatever it is, it effects a lot of similar pages.

Can you provide a few links to bugs for this?

I'd love to know the technical reasons behind that.

Read in another HN comment just the other day that the slowness was because they used some obsolete only-in-chrome shadow DOM API which "polyfilled" on all other browser, which if true sounds exactly like pulling a Microsoft.

I haven't verified it myself and I couldn't find a reference with a quick search, so while it sounds plausible, it would be great to havd a credible source to refer to.

This is not necessarily malicious though.

Probably some in-house framework they use that happens to use this old browser API. And they haven't gotten around to fixing it.

Of course it's also reasonable that nobody is pushing a fix up because it works fine on Chrome and that's what Google wants us to use.

Some kind of reference https://news.ycombinator.com/item?id=18053935

I have a Gmail account since it was invite only and I believe also there wasn't any web interface yet so I always used it through a mail client such as Sylpheed, now Claws Mail (https://www.claws-mail.org/). The very few times i had to use the web interface (always Firefox or derivatives) it worked without hitches on FF too, but a few months ago I was doing some maintenance to a customer PC (Debian+Firefox just like all my machines) and noticed how slow the Gmail web interface had become. Switching to the "old" mode helped somewhat to restore some speed, but being used to snappy dedicated mail client it still was truly slow to operate. Unfortunately today every non tech savy user thinks the browser is the internet to the point some don't even know what the word browser mean, which makes even more difficult to convince them to try anything else. Where a browser struggles with two pages of mails, I can keep available my entire archive since about 2001 on the Claws Mail client, that's over 50K mails in a dozen accounts, and query them to find whatever I need.

I had compatibility issues when I tried to use TensorBoard the other day. It doesn't even seem like they were trying for compatibility in that case.

Reply on news.ycombinator.com