Not only is IE8 a snowflake but html4 is particularly hard to get a good layout working. Lots of nontrivial layouts in pure html4 (and css 1/2) require tons of browser specific hacks.
But honestly, I can't know because he's not naming anything specific anyway, just an anecdote about how IE8 was great at being standards-compliant and safari wasn't. Which is such an extraordinary claim (and is so contrary to established precedent) that without any extraordinary evidence, I have a lot of trouble accepting it.
(And it's amazing how many people think "I'm not doing any browser hacks" while they're using something like jQuery, which is a library 100% devoted to doing browser hacks for you while abstracting you from that fact.)
I was working on top of a huge CSS framework - the majority of styling work I did was tiny tweaks to styles where the appropriate hacks had already been solved.
Also I didn't mention layout. The specific pain point was contenteditable in combination with JS - by the time I got to the code everything surrounding my little problem-space had been solved by the rest of the team. Again, specifically, Safari demonstrated wildly different behavior within that specific space, where other browsers only behaved marginally different (a relative term, contenteditable is a complete mess, sadly, still in HTML5) in a way that I could find a viable subset of functionality that worked consistently.
Well, that's the whole thing then, isn't it? The huge CSS framework probably had tons of hacks (as you say) to make things work in IE8... it's a bit disingenuous to say you wrote everything to be clean and without browser hacks, if the hacks all exist but are buried under a "huge CSS framework".
Your whole experience could probably be restated as "I was working with a CSS framework that did a bunch of cross-browser hacks for me, and the tweaks I made to it didn't work in safari", which is a pretty uninteresting statement, because it's equally likely that any issues you had in safari are the fault of the CSS, and not the browser. (In fact, much much more likely.)
Did you miss the whole of my second paragraph?
> Also I didn't mention layout
Something working in opera doesn't mean it is standards compliant.
(emphasis mine)
I use Chrome as my primary development browser and Safari and Firefox as my second and third choice. When I do get around to testing in IE 9+, it's almost an afterthought. Is it surprising then that IE is the browser that gives me the most trouble?
All browsers have quirks. It's inevitable when you have many competing implementations from companies that have different priorities. The browser you use the most is naturally the one whose quirks you will get used to.
Then I tested it in Safari. Two weeks later I gave up and basically had to `if (Safari)`. I don't know if the code still has that, but it probably does. This wasn't anything 'lacking'. Sure, it was contenteditable-related but the fact remains that I found a consistent subset in every browser except Safari.
So, yes, Safari broke my web - and it wasn't something that was missing, it was something that behaved very differently compared to other browsers. I can't remember what specifically broke it, but even ~2 years ago I had a significantly better development experience with IE than Safari with no new HTML5 features in sight.
[1]: http://help.k2.com/onlinehelp/k2smartforms/userguide/1.0.6/c...