Dear web developers, please think of the boot disk users.

Users might have to boot their computer from external bootable media such as a live USB stick, SSD, or live CD/DVD, after their operating system caught a problem that prevents it from booting.

Emergency boot media usually has earlier versions of web browsers because they are not frequently used, much less updated. Sadly, the developers of many websites have a habit of breaking compatibility for older web browsers. For example, the new audio player used by the Internet Archive (Archive.org) does not even support Firefox 57, a version that was released as recently as November 2017!

Therefore, websites should retain support for old web browsers. If not all features can be made to work, at least the essential features should work on older browser versions. Websites should not let down people who are stuck due to a computer problem. Those users should still be able to browse the Internet for help, and perhaps enjoy basic entertainment such as watching videos (YouTube, Dailymotion) and listenening to music or audio books (SoundCloud, Internet Archive) while at it.

The attached screenshot shows something no internet user wants to be "greeted" with.

Keep the Internet accessible.

how many people do these make out? 0.01%? worth the security risk and cost associated with supporting these users?

as recently as 2017....

Every time the LKML (Linux kernel mailing list) discusses removal of subsystems, the same argument comes: But someone needs it.

Yes. Someone might still use ISA. Or EISA. But this is ancient stuff. It could be removed. What is ancient, too - btw. is the average age of developers. Not gonna joke, even in NON (F)OSS companies is a severe shortness of fresh blood, either due to recent layoffs or due to too much pressure.

Plus - to get back from the example of ISA, which is really ancient, our definition of ancient needs to be corrected.

In embedded, 15 years plus can be *young*. There's a reason we have specific kernel builds and "whacky" toolchains including e.g. GCC builds (e.g. patched to cater for specific CPU instructions / embedded chipsets / ...) for that.

But outside of embedded? 1 year is young, 2-3 retirement phase, 4-5 years EOL.

5 years is the standard LTS time range.

Note that this does *NOT* mean that these 5 years are to be seen as a full life span. 2-3 is retirement phase aka planning to migrate and upgrade so one has half of the life time to get to a newer version.

If you wait till the EOL is over, congratulations - you fucked up, you now have a full rewrite at hand, certainly costing you more than a careful planned, calm migration would have cost you.

The world's moving fast. Adapt or die.

One good example for this: the megahertz race in CPUs is long over. Efficiency, be it in power or in things like SIMD extensions, are key.

The difference between e.g. an AVX enabled software stack vs an "old" SSE stack can be a game changer. Not only for specific things, but very common things like string operations.

You don't get that by sitting on your arse and saying that the migration will have to happen sometime....

You get that by adapting with careful migrations and planning ahead.

Prevention is the key. Start early. Stay calm. Make use of LTS stacks.

5 years is a long time. Don't waste it.

Or how about you just use live OS with newer version of software and browser?

Websites needing to support IE9, 10 and 11 for years overdue was horrible enough, thank god we're done with these shenanigans..

@myss Supporting Internet Explorer would be excessive, however, supporting a post-2015 (ECMAscript 6) browser should be no big deal.

@joewilliams007 Actually, many people were forced to stick to an earlier version of Kiwi Browser for Android because YouTube background playback, an essential feature for productivity, was removed at some point.

If those people were to update their browser, they would sacrifice this essential productivity feature.

@exerceo let me recommend to you: youtube revanced.

it has background play, pic in pic, dislikes, amoled, sponsorblock.

still very nieshe what ur describing.. anyhow i dislike aswell how things are stopped being supported just like that ... truly sucks

Dear USB booting users,

No.

Chilly regards,

-Web devs

Update the browsers on the live cd? It should last for atleast the session lol

@AlgoRythm This. So much This.

Also - if you are booting live from a disk on key, and not installing/rescuing a system - why not update the browser on it?

@exerceo Firefox, Vanced, NewPipe. If a feature is no longer supported in your browser your options are to switch to another one, learn to live without it, or fork it. Running an unsupported browser version is never a use case, only maybe an extremely temporary hack for a few weeks until you figure out what to switch to.

@mansur85 If a program hasn't been updated for 5 years I'm not exposing it to the internet, this is not a question of language. There's a difference between a 30 years old build and a 30 years old project.

@mansur85 Linux won the server industry because the various companies that based their business on it could contribute patches. If anything, it shows that security requires constant effort.

5 years is a bit too old in the world of computer technology. You can always find a more recent live OS.

@mansur85 I agree with your statement in the context of codebases, OP here was complaining about lack of support for 5 year old releases so I thought you're also talking about that.

@mansur85 bit of a misunderstanding...

Project life time vs project *dependency* lifetime.

If you use e.g. a QT toolkit version XY … then look at the EOL support range and make sure to migrate early.

Same for the rants boot / recovery / ... media.

A C/C++ project can be 20 years old and use the newest compiler.

It doesn't have to run on GCC 2.95 for example, just because none gave a damn.

@mansur85 then stick to it.

But don't cry that the world moves on....

... and don't expect others to keep shit running just for you. :-)

You can use an older kernel variant for example. But don't cry that the Linux kernel devs don't care for you love of old stuff.

Etc. Etc. Etc.

In company projects: Well. Bankruptcy.

I wait for the day when e.g. NPM and other registry purge old libraries to free some petabyte of data.

I'd expect that a lot of companies will be just plain dead....

Happened already a lot of times.

Or TLS certificate chains which require newer CA certificates...

...

Don't expect that legacy shit works just cause it's easier than to fix the mess one made by themselves.

I agree that only supporting the latest version of chrome is not enough. But what you are doing is highly insecure. The ERS is 102 that is the oldest Firefox you should browse the internet with provided that you updated it last April. The same goes for the other software you use please update before using.

Think about it this way. Would you bet that putting a sausage with a condom that has been expired for 4 years in a bun, is going to be an adequate barrier for sausage juices reaching the bun or the mustard reaching the sausage?
Let me remind you that if even a few microscopic drops get through you could have a huge mess.

@joewilliams007 Unfortunately, those third-party YouTube apps with background playback (in the past: FREEDi, MixerBox) also stop working some day, defeating the benefit.

They also lack tabbed browsing and a browsing history.