On 2/12/2014 15:04, me at tdiehl.org wrote: > > I guess it is hard to get it tested without making > it the default. No need to guess. There's plenty of evidence that at a certain point, software needs to be battle tested to shake the last bugs out. Take btrfs. It's been included in shipping kernels for 5 years now, yet people keep asking "...but is it stable?" Why does the question come up? Because it isn't the default filesystem. Since it isn't installed on $BIGNUM percent of all existing Linux boxes, there is room for ignorance to sprout into doubt. Firefox's PDF reader problems are certainly not all "bugs," per se. PDF is a huge bag of complex features only loosely related. I can't see Mozilla even *wanting* to implement every last behavior and feature defined by Adobe, much less accomplishing it. If your document depends on an unimplemented feature of PDF, it won't render right, so you're likely to call the viewer "broken," even if every feature Firefox's PDF reader /does/ implement is flawless. This isn't about Firefox and Adobe. It's about any software development team who's set themselves the task of following the taillights of another software development team, while the latter has a bigger revenue stream. It's an inferior strategy, if your goal is to win the race. Lots of examples of that: - octave vs Matlab - Libre/OpenOffice vs MS Office (document compatibility) - Wine/ReactOS vs Windows - SharpDevelop vs Visual Studio - Gimp vs Photoshop (PSD compatibility) If you're tempted to give "IE vs Firefox" as a counterexample, notice that I specified a bigger revenue stream. It's a necessary precondition. IE6 was leapfrogged by Firefox and Chrome because they have an independent revenue stream, while IE does not. IE also has to move around in legal leg irons that don't hobble the others. Another non-counterexample: Linux vs big iron Unix. Big iron Unix priced itself out of the segment of the market that was outgrowing the economy as a whole. Now Linux calls the dance steps.