Hacker News new | comments | show | ask | jobs | submit login
IE 11 Crashes After Clearing 5+ Input Elements (jsfiddle.net)
27 points by dahjelle 1348 days ago | hide | past | web | 62 comments | favorite



Well, I tried this in IE11 on Windows 7 x64, all three buttons works. IE11 still works.

screenshot: http://ww4.sinaimg.cn/large/798f7769gw1eceuea9cwhj211i0ox10b...


IE11 still working after clearing 20 fields

http://ww3.sinaimg.cn/large/798f7769gw1eceugu2bhlj21130r4qa4...


The crash would usually happen after the alert box in your code. Typically after one clicked in one of the input boxes and started putting in more data.


Nope, I tried to modify these fields and clicked the button more than once, there is still no crush.

BTW, what are the plugins you've installed in IE? What's the OS version? Performing more information will help everyone to find the bug.


Windows 7 Home Premium. 64-bit. I have the dynaTrace AJAX tool installed, though disabled, Java installed but disabled, the Microsoft Messenger Companion installed but disabled, and the Adobe Reader and Flash plug-ins installed and enabled.

I reset all IE settings to default, as well.

I am running in a VirtualBox virtual machine, though I've reproduced this on non-virtual machines, as well.

What else is useful?


It's too weird. I'm on Windows 7 Ultimate, 64 bit version, I've tried everything I could imagine about these textboxes, but still no crush.

Two of my friends tried on Windows 8.1 Pro x64, still no crush.


Yeah, it's like the crash is depending on something other than Windows version or IE version, but I don't have any idea what.

One commonality seemed to be Windows 7 Home Premium SP1, but I see someone else was using Windows 8 and reproduced it. Hmm…


I'm running IE 11 here and couldn't get it to crash, despite clearing the forms, focusing one of the text boxes, and beginning to enter data.


That's interesting. It crashes regularly for me. What version of IE are you running? I'm on 11.0.9600.16476 on Windows 7.


No crush, I tried everything I can imagine, including editing these controls, performing drag-drop but nothing strange happens.

ver 11.0.9600.16476

update 11.0.2

screenshot: http://ww3.sinaimg.cn/large/798f7769gw1eceujrxg30j20bf09c0tf...


Wow…same version I'm running, and on the same Windows version, but I'm regularly getting the crash. You don't suppose it has something to do with the system language or something? What other different factors could there be?


Windows 7 x86 or x64?

What plugins have you installed?

I'm using Simplified Chiense and I cannot reproduce this bug.


The only plugins I have enabled are Flash and Adobe Reader. All others—Dynatrace AJAX, Java, Messenger Companion, Windows Live ID, Blog This in Windows Live Writer—are disabled.

I should mention I'm running in a VirtualBox VM, though I've seen this on a co-workers box running Windows direct on hardware.


Here's the DxDiag from a co-worker's machine that ISN'T a VM and who can also reproduce the bug: http://pastebin.com/e6auRbR8 He has all add-ons disabled except for Java.


I've definitely come across VM bugs in the past. Perhaps this could be related to VirtualBox? IE does use a number of fancy mmu features when JITing JS that might put pressure on some of the more complex parts of a VM.


Yes, I thought it might have been that, except co-workers could easily reproduce it on their machines that aren't VMs. See the DxDiag in a sibling comment.


Very interesting. I'll have to try it on one of my home machines I installed windows on later.


Let me know what you find. Still haven't been able to figure out the pattern for why some folks have been able to reproduce this and others can't.


Following up:

This iMac doesn't reproduce (Windows 8.1 64bit, IE 11.0.2 (11.0.9600.16476), desktop and metro tested, only add-ons enabled: 1password and fiddler)


Huh. Interesting. Thank you!


Inside a VM? It's a really important information. Maybe you should try this on a real machine. Virtualization can change many things, for example, GPU-Z crushes my Windows 8.1 VM.


Sorry, I put this in a sibling comment first as HN wasn't letting me reply until now:

Here's the DxDiag from a co-worker's machine that ISN'T a VM and who can also reproduce the bug: http://pastebin.com/e6auRbR8 He has all add-ons disabled except for Java.


Here's a DxDiag report: http://pastebin.com/L96ZtQH0

I noticed one other poster who reproduced the crash was using Windows 7 Home Premium SP1, as well. Perhaps it doesn't work on other Windows 7 versions?


Nope. No crashes for me either. I'm also running 11.0.9600.16476


IE 11.0.9600.16384 on Windows 8.1


I have Windows 8.1 (x64), and my IE version is 11.0.9600.16476. I got it to crash quite quickly, and it's repeatable.

I typed text in all 5 boxes, pressed the first button, started typing into the first box again. Bang: "Internet explorer has stopped working". pressing "close" on that modal dialog just reloads the tab though.


No crash for me either. v 11.0.9600.16384


This sounds very much like the bug at http://connect.microsoft.com/IE/feedback/details/808033/ie11...

The (surprising) solution is to first assign a space character, then the empty string.

So (I'm guessing), in that http://jsfiddle.net/dahjelle/zVjfY/ code, change

field.value = clearValue;

to

field.value = ' '; field.value = clearValue;


Fascinating! Yes, indeed, that's a valid workaround!

Not exactly pretty, still, but better that that weird setTimeout mess.

Thanks for the pointer!


Well, shit. And for a few short years, it seemed that IE has outgrown its monstrous reputation it had rightfully acquired in the 00s; nope, still the usual "spend enormous time catering to my quirks" IE. This makes me sad.


It hasn't outgrown it's monstrous reputation and it will probably never. IE has long gaps between updates so it will always quickly fall behind. Software can still install toolbars and change your default search engine. IE is not open source. IE 11 does not implement X standards for business reasons, where X is WebRTC, some missing CSS thing, touch events, etc. You can't test for IE except on Windows. IE's developer tools are terrible. IE is used by people bad with computers and corporations with bad IT departments.

For me to care about IE, it would:

have to be available on Mac OS X.

have to have bigger market share in mobile

just have more people using it in general (because frankly it's not used much by any of the people we care about).

have much, much better development tools

Our development cycle is to develop in Chrome Canary, and proactively test and fix for Firefox. We don't test in IE but fix IE issues when someone complains about something not working in IE and if we have the time or think it's important enough.


IE11 has touch events, they're just a different api /than everyone else/.


I have come to terms with this - since most of the compatibility issues can be papered over nowadays, and there are viable alternatives - all I was hoping for was "write and debug once, give it a quick look in A-grade browsers, fix minor quirks", without the usual "...and then spend double the time fixing IE's idiosyncrasies."


Many of these points are partial truths or at least not as damning as it might seem.

IE has had long gaps in the past because of IE being tied to larger updates and many people having invested in old technology depending on them changing that slowly to allow time to change. The IE team at MicroSoft is clear that they intend to do more regular releases and push updates out automatically. While it won't be monthly like other vendors, it's certainly a huge change from where it was.

Software still installs toolbars on Chrome too. I recently tried Windows again for the first time in about 13 years and I'll tell you that I was caught off guard. Toolbars ended up in Chrome AND IE. Not sure if this is a browser problem as much as a software installation problem. Wish I would have learned about Chocolately earlier.

Open source isn't something that makes the browser better at being a browser. It's possibly important in other aspects but you should be complaining about other browsers then. Many include lots of proprietary code on top of open engines like WebKit (i.e. Safari is really not all that open source) and then custom engines like Opera mobile.

WebRTC isn't even half baked yet. Complaining that they don't release it to millions of devices as a feature is silly. It's great to see browsers experiment with new APIs that become standards... which leads me to: touch events.

Touch events were a proprietary outgrowth of the early mobile web. While they've become defacto for mobiles, it's a limiting API that doesn't really help unify the way we handle input at all. MSFT released a vendor extension called pointer events which is a much better API that has now been proposed as a standard. It handles touch, mouse pointers, pen digitizers, and has space to interpret more kinds of devices. Much better than where touch events left us. It's sad that more developers don't demand this unified API either since it makes it far easier to implement cross-input-device friendly content.

I haven't spent much time with the IE developer tools but it seems that MicroSoft now provides free VMs that can be installed to run IE for development (http://www.modern.ie/en-us/virtualization-tools) and their developer tools seem to have been improved (most people now just complain they want chrome's developer tools everywhere).

So what is left is that I see people either stuck in the past or thinking that this whole thing is IE vs everyone else. IE's got the same problems as many other browsers. We just pick our check list with bias by telling all browser vendors they should just become Chrome. What's the point in an "open" ecosystem if people just demand we all do things exactly one way defined largely by one vendor which happens to develop these "standards" as proprietary extensions just like IE has for all sorts of things (XHR anyone?).

>_<


> We just pick our check list with bias by telling all browser vendors they should just become Chrome.

Now this is the scary part - seeing "only for Chrome" everywhere just brings back those memories of seeing "only for IE". Another browser war is looming on the horizon.


strmpnk, what the fuck?

> Software still installs toolbars on Chrome too.

Chrome does not have toolbars. You cannot build a toolbar with Chrome, the browser has no extensions API to build such a thing. You also cannot programmatically install extensions. Windows software cannot install extensions into Chrome, barring some kind of malicious exploit or vulnerability in Chrome, Chrome will ignore them. They have to be installed by the user from the Google Web Store or via the developer tools, which has to be done manually.

I'd like to hear if anything I've said in the previous paragraph is wrong, I have never seen a toolbar in Chrome ever. You can't even build it into the website with scripts because of how those APIs work and anyway websites could detect that and I've never seen it.

> WebRTC isn't even half baked yet. Complaining that they don't release it to millions of devices as a feature is silly. It's great to see browsers experiment with new APIs that become standards... which leads me to: touch events.

This is bullshit, as we've built WebRTC into our application and it works perfectly in Chrome and Firefox. In addition there are entire businesses based around WebRTC availability so, you are mistaken here.

> Open source isn't something that makes the browser better at being a browser.

It matters to developers who build websites. I can debug my application all the way into the Chrome source, and have had to when JIT ruined my JavaScript.

Even 6 month updates are way too long, and I doubt that we'll see updates in that time frame anyway. The IE VM tools are slow and horrible, it's not usable, I've tried it. It's no replacement for native development and we're not going to support it. Maybe some draconian soulless corporation has made browser vm's an option to their poor mac engineers, but fuck if I'm going to do that. Your entire apologist response is factually incorrect or misleading. It's very misinformed and one-sided (although I also come across one-sided). Weird post anyway.


belluchan, slow down a little before you assume I'm just making shit up.

Toolbars as in default search and browser plugins which do layer on buttons on the UI. I'd count that as a toolbar. Happened to me just last week when I installed software on Windows 8.1. Chrome definitely is just as prone to infected additions. Count me as surprised.

Re: WebRTC, how long has that actually worked? WebRTC is still under development. It if worked in the last few months chances are it didn't work the same way before. Chances are it won't keep working that way later. It's exciting tech, I love WebRTC but calling it a missing feature before it's landed for a few months is kind of asking for a bit much, even for Firefox and Chrome (audio APIs, MathML, JavaScript language features, &c). The fact these two browsers interoperate at all is because of a ton of hard work they put into collaborating to make things work. Of course, "work" being defined by what is easiest for their current implementations to change. Still, you probably assumed I was trying to bash WebRTC, I'm not but it's very new tech in terms of the current interoperable feature set. Give it six months and then come back to complain.

I'm glad you solved your Chrome issues with tracing C++ code. Sucks when that happens.

I don't see how this makes supporting IE entirely impossible and super painful. You have so many fewer variables in play with the IE release cycle that it's hard to complain. I've had Chrome bugs that triggered on only certain builds on certain operating systems that took me a ton of work to reproduce. Looks like the problem reported here is not hitting everyone either but it's already narrowed down to only a few possible versions and operating systems.

To make things worse, Chrome didn't have much help on debugging on mobile devices for a long time. It was quite hard to get an accurate rendering and runtimes with complete debugging.

The IE VM isn't slow here. I just downloaded and tried it to write this comment. I get WebGL and all running in my VM. Not sure what you're talking about. Felt fast to me.

Anyway... I'm not an apologist. I build software and honestly most of these rants I see are complete FUD. They show either ignorance about what options are available or the bias of choosing one flavor of proprietary over another. I don't care if you use or support IE or not (I don't use IE nor do I run Windows on anything but one machine to play some old games I like).


With each release of IE since IE7, there has been some kind of marketing push saying "THIS version is the one that is modern and totally standards-compliant, really", and a bunch of people repeat that. Then it turns out to be buggy. Then the cycle repeats. No one should be surprised at IE11's flaws.

I believe IE's quality has been fairly consistent over the years. IE6 had a monstrous reputation mainly because it was outdated and widespread for so long.


I have passionately hated IE6, 7, and 8; IE9 has managed to reduce this to a resigned "meh, good enough, what took you so long?" (which _was_ a significant change to the better). I was hoping that IE was finally shaping up to be an A-grade browser. Oh well.


There's definitely still some weird IE bugs, although they're more (but not exclusively) on the JavaScript side now rather than CSS.

For example, if you have a form without a submit button in IE 10, and a `<button>` element anywhere else on the page, IE will treat the <button> as the submit and "click" it for you rather than submit the form. If you don't have a `<button>` element though it will submit the form as expected.


There are various inconsistencies between browsers (including between version of IE) when you do not explicitly set the type attribute of a button tag. Some will default to it being a submit button and some will default it to being a "button" button. In the instances where we hit this problem making sure to specify a type seems to make things more consistent.


I think this is a long time bug, not IE10 specific.


Tried this and it crashed for me on IE 11.0.9600.16476 (Update Versions: 11.0.2) on Windows 7.

Screenshot: http://i.imgur.com/oSi1rlL.png

Steps:

- Filled in all fields

- Clicked 1st clear button

- Clicked in 1st field

- Tried to type something


x86 or x64?

-- update --

I tried the identical reproduction process and IE11 still working.


Windows 7 x64 (Home Premium SP1)


On a real machine OR inside a VM?


This is on my actual machine, not a VM.


We ran into this in out application the other day, and it's a surprisingly "normal" thing to do to result in a hard crash of a browser.

I'm especially curious if anyone knows:

1. Is there any hope of this being fixed in the near-term? The MS bug report didn't have much information. I'd sure like to avoid implementing the somewhat crazy workaround if I can help it.

2. It seems like this ought to be a pretty widespread problem. Are others seeing it? Are there other workarounds that aren't so ugly?


It doesn't work here :-( Win 8.1, IE 11.0.9431.228


Reports of no crash are actually more interesting to me than the crash at the moment…maybe I can protect our customers. Are you running Windows 7 or 8? Any add-ons?


I'm running on Windows 7 (x86_64) with IE 11.0.9431.0, No addons at all and there are no crashes.

DxDiag: http://pastebin.com/M0msHVh3


So it is a bug in earlier version? or a Windows 8.1 specific bug?

BTW, what are the plugins you installed in IE11?


Win 8.1 Pro, no plugins


I was also able to reproduce this on all of the IE 11 instances on http://browserstack.com: Windows 8.1 desktop and metro and Windows 7.

I did notice that the crash very occasionally didn't happen the first time, but did, perhaps, the second.


personally, i think it's adorable that microsoft keeps trying to make a web browser.


It's also frustrating as hell that they finally improved the dev tools and then removed all the browser and document modes. Pretty typical though for them. One step forward, three steps back.


Noting will ever pull me away from Chrome's DevTools! Not IE, not even Firebug.


I feel the same with Apple's Safari. If you're making a browser and its platform specific, the internet would be a better place if you'd put it out to pasture.


To the best of my knowledge Safari is available for OS X, Windows, and iOS.

I used to use it on Windows because I liked the font rendering more than Chrome, Firefox, Opera, or IE.


Safari for Windows was discontinued after version 5.1.7.


In case anyone is wondering, it doesn't crash in IE 8! :-D


It didn't crash for me.




Applications are open for YC Winter 2018

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact

Search: