Google is considering removing alert() | Hacker News Reader

Google is considering removing alert()

twitter.com

43 pointsjoshdance39 comments

https://twitter.com/estark37/status/1422694856544059396

> as a developer it's good practice to test against early release channels of major browsers to learn about any compatibility issues upfront

This is a rather alarming viewpoint which I think highlights Chrome's quest to become the One True Implementation of web standards. In addition to the major browser vendors I'm now supposed to be testing against beta and nightly builds in order to ensure that my website is not going to break because Google decided to eliminate something? Yikes.

Contrary to what seems to be Google's belief, most websites are _not_ continuously deployed SPAs with a team of engineers who work on each release. The indifference shown to actual developers here is staggering.

Well said. It's stunning how ignorant Google is about the web. The vast majority of the web is old and poorly maintained or not maintained at all. It relies on things not breaking, web tech being backwards compatible.

This assumption that behind every website is a team of developers maintaining it for its entire lifecycle is a stubborn and elitist Google fantasy.

They're imperious, but never ignorant. A website without a team of maintainers is beneath their concern. What are you going to do, fax them a nastygram about it?

You're right, we can't do much about it. Even if a Google change harms a million people, they consider it peanuts.

Yet..."arrogance precedes the fall".

One day they will pay the price for their carelessness and harm, and I welcome that day.

Good. Alert is an archaic mechanism with tons of useful alternatives that don't involve blocking the DOM. For example, console log for debugging/demoing.

On the other hand, I'm seriously worried that Google is using its position to force large changes to the web.

it's not good, this change will break huge swathes of the web if it goes ahead as planned, lots of older websites just won't work any more and many newer ones will also be affected.

There is no place for an out of dom signalling mechanism like alert() on the modern web. I would hope to see a multi-year phase out of something this integral. I can't remember the last time I saw it used for something useful.

deprecation is totally fine, log a warning in the console, discourage people from using it, but don't throw the existing web under a bus because your UX team can't think of a nice way to visually distinguish it from the browser's chrome!

> can't think of a nice way to visually distinguish it from the browser's chrome

i don't think it's about visual design. Having a signalling mechanism that lives outside of the dom is confusing to users, and alert is a rare exception from wayyyy back when. It looks like google is doing this for user security, and I support it.

I don't support the way they're going about it. If Mozilla and Microsoft and W3C are willing to get on board, I would greatly support it. But this is more throwing their weight around, and it sucks!

>existing web

i'm gonna be honest, i havent seen a meaningful alert be used in recent memory, and i work with a lot of web apps. All alert()s could be easily replaced with something inline. I'm not a huge fan of breaking changes, but this is one (with a multi year rollout) that I would support.

It isn't uncommon for me to find a site that uses it legitimately. Sometimes for telling the user info, sometimes for reporting an error, and sometimes for other goals. Sure, it'd be better if they did it in a more normal manner (writing their message into an HTML paragraph), but they do use it. But a lot of these sites are dead (though some aren't) and so it for the most part doesn't matter if you remove it now or after three years. They'll still break or lose functionality. Just spit out a warning to the dev console when it is used.

>i'm gonna be honest, i havent seen a meaningful alert be used in recent memory, and i work with a lot of web apps.

I see them used very often to indicate unsaved changes when a user leaves a page.

Isnt that confirm()

> In general all of the simple dialogs (alert(), prompt(), confirm(), and beforeunload) are deprecated and being removed slowly but surely from the web platform. They use trusted browser UI, which opens them up to abuse, and they block the event loop, which is not in keeping with the web's cooperative task model. So this is just a first step toward eventually completely removing them, which will take many years.

https://github.com/whatwg/html/issues/6897#issuecomment-8857...

that's confirm(). alert() cant block a page close in progress (by design, afaik)

I'm actually not a huge fan of the design of confirm() either, but at least that has a valid, modern use

The UX issue is directly highlighted in their rationale for the change.

Calls to alert(), confirm() and prompt() cannot be polyfilled or safely replaced in the general case because they explicitly block the main thread. That behaviour might not be desirable, but it was the path chosen when the relevant code was written. While some maintained applications will be able to switch to asynchronous alternatives, many older and unmaintained websites - the long tail of the web, will stop working.

Pretty sure it's a different problem - alert() is a true modal dialog that blocks almost all other normal interactions and is very disruptive. Back in the day when apps weren't constantly gaming all user interactions to squeeze on extra ad click out or whatever, it was still abused - but rarely.

Now if you wander down the wrong internet 'alley' you'll get mugged by someone using alert() to almost completely hijack your normal session. Especially for older folks (my grandpa really suffered from this), it becomes impossible - to the point they'll even call someone up and give them their credit card number sometimes so they can 'get to their email'.

Warnings to the dev console won't stop this type of harassment of users - only stopping it from being the modal dialog it is supposed to be (per alert() prior behavior) will.

Every major web browser has implemented a "do not show alerts from this site anymore" checkbox that appears after 2-3 consecutive alerts.

"Protecting the users" is an absolute lie.

A good compromise could be to have it trigger a non-blocking pop-up notification within the page, since the blocking is the main issue

Heck, if they did that it might come back into style. Pop-up notifications are a pain to implement from scratch

>Alert is an archaic mechanism with tons of useful alternatives that don't involve blocking the DOM.

event loop, but the _whole point_ of Alert is being MODAL and BLOCKING. If I didnt care about that I would just pop fixed position div. Its meant to pause whole page, thats its designed role.

I seem to remember browsers (Netscape or Opera?) around 1999 having "Never show this dialog again" checkboxes meant to prevent alert() spam.

I've rarely seen alert() used in the past few years outside of as a means of harassing users on malware sites (have you ever accidentally clicked on a malicious ad on a mobile phone and struggled to escape the site due to alert() spam blocking interaction? It's terrifying as an end-user). This seems like a more than reasonable decision given an adequate phase-out period.

And what happens when Google makes an unreasonable decision about what should be a part of the web standards and what should not? Who even has the capability of making them reverse course at this point?

No. They are not. They are considering disabling it for cross-origin alerts from iframes.

https://twitter.com/estark37/status/1422694845076762629?s=21

Everything else is speculation. It may even be likely speculation! But it is speculation nonetheless.

Yes they are, and within a few months, Google will continue fucking over the web. They are explicitly saying there will be breakage because "wah wah it's on the main thread".

https://twitter.com/estark37/status/1422694855390629893

It would be speculation to say "Google is removing alert()"

It is not speculation to say "Google is considering removing alert()". Multiple engineers from Chrome have confirmed on Twitter that they have an eventual goal of removing alert() at some point in the future (or making it non-blocking and removing prompt/confirm, which is the same thing). See https://twitr.gq/domenic/status/1422647331804037120 as one example.

Maybe it won't happen! There's a lot of time for feedback. But it is accurate to say that Google is considering it.

Also as a small sidenote, they are not considering removing alert() for cross-origin iframes. They are removing it for cross-origin iframes and have very temporarily walked the changes back in order to give devs more time to prepare. Short of some kind of massive blowback that forces the devs to change their mind, Chrome's decision to remove alert() from cross-origin iframes has already been finalized.

Isn't "don't break the web" supposed to be the most important priority?

They're talking about removing alert(), but there's no mention of confirm()?

I believe this tweet has been twisted from the original tweet, which is that _cross origin_ alert (e.g. from an iframe) is the thing being removed. Using alert etc from the main frame will continue to work.

They are planning to deprecate them from the platform entirely, this is the first step

Please provide a link, or stop passing speculation off as truth.

Tweets from Google employees who work on chrome:

https://twitter.com/estark37/status/1422694855390629893

https://twitter.com/domenic/status/1422647331804037120

https://twitter.com/domenic/status/1422647483944034308

https://twitter.com/domenic/status/1422704005453582337

Those links contradict your assertion, instead indicating a move to "non-blocking alert()"

That breaks the web by definition. There is no way to implement a non-blocking alert(), confirm(), or prompt() in a way that doesn't break existing code that uses them, and that's a lot of code, a lot of the web.

It can block JavaScript execution on the page for compatibility with existing code, while leaving normal browsing unaffected otherwise.

the alert(), confirm(), and prompt() themselves interrupt normal browsing, so I'm not sure what you mean here. They're synchronous so block the main thread, but they also block other user interactions until they're dismissed, so what would be the benefit in leaving normal browsing unaffected if the user cannot browse anyway?

From those links:

> Yeah, I think the most likely eventual "removal" path is actually non-blocking alert(). (confirm() and prompt() cannot be converted to non-blocking though, so they'd have to be fully removed)

Blocking Javascript execution is a fundamental part of the `alert` spec. Replacing it with a different, non-JS-blocking thing that happens to share the same name doesn't mean that `alert` wouldn't be removed as it exists today.

And there's just no way at all to make `prompt`/`confirm` non-blocking, they have to block the main JS thread in order to keep the same API.

We tried this topic yesterday on the FP with a link to a bug tracker, and it went quite badly. I wish someone would write and post to HN a serious full-length discussion about what's going on, with clearly separated "Google's stated intentions as of today" from "Reading the tea leaves" from "This is outrageous".

I still don't even fully understand the problem, because everyone just defaults to linking a bug tracker that's of lower quality than a Reddit discussion, and I can't trust HN comments to explain it because of all the outrage and assumption-by-framing. Ugh.

From cross origin iframes. This is a good thing IMO.

there doesn't seem to be a way to opt back into it though. At the very least they should do that

Reply on news.ycombinator.com