For example, some common methods are:
* scrolling with a mouse by clicking on a traditional scrollbar
(this is what is being reimplemented)
* using a scroll wheel or trackpad drag
(compatibility is *often* added for this, but not always)
* dragging touchscreen (support varies, getting better)
* using the arrow keys (sometimes supported)
..and a few of that almost always break when scrolling isn't left to the user agent: * PageUp/PageDn
* some other key to page down, such as the space bar
* a browser extension to auto-scroll the page slowly
* accessibility tools of various types
* no javascript support
* the foot pedal I made out of MIDI keyboard sustain pedals and an
old USB keyboard that is mapped to a non-standard X KeySym and
bound as an alias to PageDn in firefox (it's Meta in emacs)
* the scrolling UI for next year's Cool New Gadget that we cannot
support because the device doesn't exist yet, but the manufacturers
of the device can add support for standard HTML/DOM features
Also, even in the case where the UI method is guess correctly, the details may be different. For example, I set firefox's mousewheel.default.delta_multiplier_y to 500 so each scroll wheel click is a much larger movement of the page, because my mouse wheel is not a mouse wheel - I use thumb buttons for scrolling because the normal wheel was causing RSI problems.The basic problem is a common misconception about what HTML, where the client (via their User Agent) controls the presentation. As they are in the hands of the users, there will always be a lot of variation as those users tailor their tools to their personal needs. The only way to maintain any particular presentation feature across all clients is to turn the WWW back into TV. The alternative is learning what can be specified by the page (pixel scroll position) and what is NOT in the spec (how the user and their User Agent interact), while remembering that any of these features may not be implemented the way you think they are.
Though we could discuss it all night long, in this particular case, people seem to be complaining about a chrome bug that doesn't let you scroll at all after you've done certain things ( can't really find steps to reproduce).
Which brings us again to the dangers of playing with scroll.
Chrome on OSX and chrome on android are quite the majority (and sometimes over represented on web-devs) so maybe that's why you don't see much issues.
This issue in particular happens to me with Chrome in Ubuntu.
Very frustrating indeed on FF on a desktop.
Is there some other way of scrolling that I haven't thought of, and therefore am not using day-to-day, that breaks for other people, or do I just put up with it unconciously, switching between methods 1-4 automatically on the laptop, and never notice? Or, are people using iPads or iPhones, which I do not own, and they are more susceptible? Or Browsers-other-than-Chrome, which I guess must exist ;)
Genuinely curious!