Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
×
Chrome Firefox

Firefox and Chrome Versions '100' May Break Some Websites (engadget.com) 92

As both the Chrome and Firefox browsers approach their 100th versions, what should be a reason for the developers to celebrate could turn into a bit of a mess. From a report: It turns out that much like the Y2K bug, the triple-digit release numbers coded in the browsers' User-Agents (UAs) could cause issues with a small number of sites, Bleeping Computer reported. Mozilla launched an experiment last year to see if version number 100 would affect sites, and it just released a blogpost with the results. It did affect a small number of sites (some very big ones, though) that couldn't parse a user-agent string containing a three-digit number. Notable ones still affected included HBO Go, Bethesda and Yahoo, according to a tracking site. The bugs include "browser not supported" messages, site rendering issues, parsing failures, 403 errors and so on.
This discussion has been archived. No new comments can be posted.

Firefox and Chrome Versions '100' May Break Some Websites

Comments Filter:
  • Subject? Really? (Score:5, Insightful)

    by Trip Ericson ( 864747 ) on Thursday February 17, 2022 @09:50AM (#62276305) Homepage

    Crazy idea--built standards-compliant webpages and you won't have to care about browser version numbers at all.

    • Re:Subject? Really? (Score:5, Informative)

      by jellomizer ( 103300 ) on Thursday February 17, 2022 @10:04AM (#62276357)

      Currently checking with HTMLTest.com [html5test.com]
      My Version of Firefox has a score of 512 out of 555
      My Version of Chrome has a score of 528 out of 555
      My Version of Edge has a score of 528 out of 555
      If I were to use a different OS with the updated version of these browsers I would have different numbers,

      Last year these numbers were lower, So even with the top three major browsers there is a difference in what standards are supported and what works or not. Old versions may not support common things now, so you might still need to check for compatibility. Especially now that some sites use to rely on flash now have to do HTML5 and recode everything.

      • Can't developers do something like HTMLTest.com does though and check for specific features that are needed rather than rely on a user agent string? Otherwise you need to keep up to date with several browsers (not all of which are available on all platforms).
        • This is what competent developers do. I've always had this policy for any sort of "feature" that falls outside straight tags. Check to see if it works. There's almost always a way to do that. Anybody using browser or version number to make the decision on whether or not the page will render correctly, without giving any alternative? They've failed.

        • by Merk42 ( 1906718 )
          It depends, but usually yes they can and should do feature detection.
          There are some rare instances where a browser "says" it supports something, but it really doesn't or it's an older syntax.
      • Comment removed (Score:5, Interesting)

        by account_deleted ( 4530225 ) on Thursday February 17, 2022 @10:34AM (#62276485)
        Comment removed based on user account deletion
      • Re:Subject? Really? (Score:5, Informative)

        by Moridineas ( 213502 ) on Thursday February 17, 2022 @11:39AM (#62276703) Journal

        Interesting.

        Safari 15.2 on older version of macOS, 515 out of 555. A lot of of missing codecs, a few mostly mobile-specific features, not many major features at all. (I didn't even know there was an <input type="month"> element.)

        iOS 15.3 Safari = 464 out of 555

      • That doesn't even catch everything. I had to do a "basic" website. And Safari and Chromium count where the 0 pixel is differently if you hide the nav bar for a full screen experience. There were a bunch of other differences. In other words, two separate code paths.

        Safari is now the largest impediment to a standardized web design. Apple really is the new Microsoft.

      • it breaks something every time, on an offline pc i get about 100 telemetry errors (from firefox ! not windows)
        where IS the linux community when you need one (that works?) - o thats right
        ON CROTHUB ! (of all places)
      • It might help if Goole at all would stop pumping out new standards every six months, so that web browsers can focus on properly supporting what's already there.

    • by Megane ( 129182 )
      That sounds nice, but when I forced a higher Gecko version number in the user agent string of an older Mozilla browser, Google Maps actually crashed the browser almost immediately, because some features were "implemented", just not successfully. (This was about two years ago when Google turned the strictness of captcha's browser check up to 11. I thought it was caring about version numbers, but it was actually freaking out at garbage collection pauses in the stream of live mouse movement data.)
    • Yup, the more we have standards, the more that people ignore standards..

    • Video sites lock people out according to DRM levels. Like Peacock forbids Linux. The browser version may be relevant to determining compliance with restrictions.

  • If it breaks the website, because the webdeveloper/coder couldn't deal with numbers doesn't deserve to be visited.
    Javascript is way too lax converting string and numbers anyways, if the developer found a way to break that even further then they deserve what comes to them.

    • by AmiMoJo ( 196126 )

      Just wait until Windows hits version 89, then we are in for some trouble.

      • At a rate of a major windows version every 3.75 years that is going to take 290 years. Unless Microsoft decides to uses Googles Version Numbers.

        • They ought to use mythtv versioning. After 20 years they're still on version 0.28. My projects start on version 0.0.1 and every major commit adds 0.0.1, and every bug fix release adds 0.0.0.1.

          • by Mal-2 ( 675116 )

            I just call everything an Alpha at first, then a Beta, then a Gamma, then a Delta... who decided that there's alpha, beta, and release?

            Also who decided that of all the kinds of shit out there, bat shit is the craziest?

          • According to the MythTV website [mythtv.org], their current version number (released in March 2020) is 31.0.
        • It's more likely than you might think. As a matter of fact, some time ago Windows had made it all the way up to version 98 before Microsoft changed their numbering scheme.

          • by pacinpm ( 631330 )

            It's more likely than you might think. As a matter of fact, some time ago Windows had made it all the way up to version 98 before Microsoft changed their numbering scheme.

            I think I've also seen version 2000.

      • Wait until the version numbers become emojis.
    • by robsku ( 1381635 )

      All it takes is to stupidly write a regex that assumes 1 or 2 digits for version number. It's dumb, yes, but many good programmers sometimes make stupid assumptions about stuff like this - usually happens when you're highly skilled and writing parts of code you might already feel you've written gazillion times in your life, one might go kinda on autodrive and then - you write stupid shit like this; I know it's happened to me, but this specific case...?

      Well, I'm not certain there was never a time when I migh

    • I like the list of the websites affected. It's like a whos who of bad tech companies.

  • by bradley13 ( 1118935 ) on Thursday February 17, 2022 @09:56AM (#62276323) Homepage
    This is the kind of mistake I expect students in programming 1 or 2 to make. A professional programmer? Well, I suppose for some definitions of "professional". /s
    • Back in my consulting days.
      A potential customer, I want a web site that does this and that.
      I quote them the price to make a site that does this and that.
      That is way too high, I won't go with you, but Ill just ask my kid nephew to do the work. He can code HTML too, why am I wasting my effort with you guys who does kids work.

      The thing is, coding isn't that hard, anyone can do it. just like nearly anyone can draw, but the cost is in the details. You can get it done for cheap, it may be passable, but if you ha

    • by ddtmm ( 549094 )

      This is the kind of mistake I expect students in programming 1 or 2 to make.

      Or programming 001 or 002.

    • by nagora ( 177841 )

      This is the kind of mistake I expect students in programming 1 or 2 to make. A professional programmer? Well, I suppose for some definitions of "professional". /s

      Which mistake do you mean - the websites' or the people counting every fucking release of their browser with a major number because "big numbers are better"?

    • The definition of professional is getting paid to do it. Employers are lowering their standards.

    • I tend to validate and run sanity checks on every input I can, specifically to make sure it cannot break.

      While I agree with you that a weird version number should not break a website, but we need to know how it is breaking to decide if it is bad programming.

      First off, as of a few years ago it would of been perfectly reasonable to believe that no sane person would ever reach a 3 digit version number. As it turns out we are now being forced to create websites that cater to the insane.

      In certain extreme cases

    • This is the kind of mistake I expect students in programming 1 or 2 to make. A professional programmer? Well, I suppose for some definitions of "professional". /s

      To be fair, I bet some of these sites were expecting us to be up to Firefox version 7 in 2023, and maybe hit double digits sometime before 2030. It wasn't the programmer's fault that Mozilla, Google, and the rest of the world decided to go full retard with version numbers, advancing the major version for even the most minor of change.

      Programming 101 teaches you to only update major version numbers for major changes.

  • who cares (Score:5, Funny)

    by usu4rio ( 1115041 ) on Thursday February 17, 2022 @09:57AM (#62276327)

    As long as slashdot unicode works fine, who cares?

  • by luvirini ( 753157 ) on Thursday February 17, 2022 @09:57AM (#62276331)

    .. want to decode the browser for some silly reason and present all those "your browser is not supported" messages.

    Not supported should be on the user, not the website.

    And yes I know, if you want to use some cutting edge things they are different in the different browsers like some storage api things and similar, but for 99%+ of websites there is no need for that level really.

    • Maybe I have been out of the web design and coding game for too long, but I haven't used browser detection since the bad-old days of IE 5/6.

      Back in the day I had a website specifically for IE, one for the Sega Dreamcast web browser (it was 1999 and I had one so why not, I'm a nerd), and one for everyone else. Nothing seemed to break on these browsers back in the day since I ran basic stuff (some basic javascript, some early CSS, images, basic HTML with tables and eventually DIV layers). It was a small websi

  • by Ecuador ( 740021 ) on Thursday February 17, 2022 @10:02AM (#62276353) Homepage

    Two things are stupid here: A website parsing the UA agent string in "clever" ways and breaking if it fails, is a very stupid idea, given the fact that there are no standards around that. Then, increasing the major version number by one every few months is another stupid idea, why suddenly go against decades of industry practice giving users something that is quite unhelpful? No idea how far v68 is from v71, how recent they might be compared to the current latest etc.

    • If the release schedule is consistent the difference between v68 and v71 is like one thing that is pretty easily deduced. They are at least 3 x the release window apart.
    • by Bahbus ( 1180627 )

      No, just the first is stupid. Most of them aren't even "clever", they're just bad parses that happened to mostly work. And the failsafe (on the website's behalf, not the browser) should be 100% standards compliant version of the website, if the parse fails or userAgent is otherwise not identifiable. Then, and only then, is it then on the browser. It's really not hard to do, these websites devs are just lazy and/or uneducated.

  • It's Windows 9 all over again!
    • That's why I stayed on Windows 9. Didn't want to deal with all the bugs Windows 10 would cause.
    • Microsoft has absolutely gone out of their way to break version checking methods. Nothing works consistently from early Windows through 11. Mostly recently, they broke the ability to get the version of kernel32.dll and other DLLs, by rewriting the GetFileVersionInfo API to smash together fixed version info from a .mui file somewhere, with the string file version, which still comes right from the DLL you're looking at. So on Windows 10 (newer versions), you get version 6.2 (Windows 8, which makes no god damn
  • hbo max needs make Linux work with there drm!

    • by Megane ( 129182 )
      It's not a problem for me, they have nothing that I want to watch.
    • by Lando ( 9348 )

      In the past HBO MAX was blocking Linux, but I haven't had any problems with it for over a year. What browser are you using that doesn't work on HBO Max?

  • Good. Hopefully this will teach web developers not to depend on User-Agent strings. Or to sanitize their inputs. Or to write code that gracefully degrades when it sees something it doesn't understand.

    Yeah, yeah, I'm dreaming, I know. Absolute best case will be that sites change a regex from "(\d\d)" to "(\d\d\d?)". Well, those that don't just have a big lookup table with every User-Agent string ever seen, that is.

  • Does anyone have a list of the version strings used so far ? Have they been consistent. Can the parsing libraries determine which is newer: 1.201 or 1.20.1
  • by jddj ( 1085169 ) on Thursday February 17, 2022 @11:24AM (#62276659) Journal

    You're not supposed to be sniffing User-agent for site configuration.

    If you feel you MUST, vs., you know, actually building a good standards-compliant web site, you instead test individual capabilities before you use 'em.

  • by williamyf ( 227051 ) on Thursday February 17, 2022 @11:31AM (#62276681)

    It will be major #91 until about sept this year. Longer than regular channel.

    Let the regular channel gamma-test this issue for you.

  • Parse a three digit number? Computers can do that? Apparently not all.
  • ... poor coding of the websites it what will do the breaking.
  • Some applications had problems when MacOS 10.4.10 and 10.4.11 were released. These apps compared version strings with a plain ASCII string comparison, which meant "Version 10.4.10" was less than "Version 10.4.2".

    And MacOS had a numeric string comparison option for many years, which makes it particularly annoying. They also had a function that splits the version string into three numbers 10, 4 and 10 which would have been easy enough to compare.
    • by spitzak ( 4019 )

      The function to use strverscmp [apple.com].
      This also has the huge advantage that it avoids "locale" stuff, which is causing "foo.jpg" to sort after "fooa.jpg" (for one of many many annoyances).

  • Could they (the developers) just name the new versions 101, etc?

    My two bits

  • Go from S9, S10, to S20, then to S21, S22. Qualcomm took it a step further and went from Snapdragon 855, 865, 888 (a Terminator nod/8 is a lucky number?) to "8 series" Gen 1 for their latest SOC.

    Firefox could call version 100 something like Series 10, or the 10 squared series.
    "S10.0" then "S10.1", and so on, until they arrive at "S11.0".

  • A better one is

    "Firefox and Chome version '100' may fail to load some web pages properly."

    The problem isn't that the web sites have something wrong with them, its Firefox and Chrome that are broken by upping their version numbers faster than someone who has chronic diarrhea goes through toilet paper.

    • by Bahbus ( 1180627 )

      Which is inherently the website's problem not Chrome's or Firefox's. If someone hardcoded the parse of a number that quite clearly increments over time, then they're a fucking dumbass. Just like the retards that used to program with 2-digit dates and then were like "oh fuck, the year 2000!". It doesn't matter if Chrome increments its major version daily or yearly - it's not static, not meant to stay static, nor does it decrease. Those 3 things alone tell me I should be dynamically parsing that string if I'm

  • I'm going to build a new browser that is 100% standard compliant and if it comes across a website that doesn't load for some reason, it just gives a big button that automatically emails/tweets/instas/etc the website owners telling them that their dogshit website is broken. That's it. No failsafes. No "we'll try a different userAgent". Nah. That shit is tired, boring, and fucking stupid. Don't compensate for other's shittiness - that's how we got to where we are. Shame them and force it to stay broken until

I have ways of making money that you know nothing of. -- John D. Rockefeller

Working...