Your last two posts seem to paint opposite pictures of the software industry.
The first painted them as fastidious and enthusiastic, but the latter as slack a-holes. So which is it?
It is more that most are somewhere in between. Certainly in my work with rolling out a testing methodology to a major retail store chain, and improving the governance of projects to ensure testing and documentation were adequately funded at a major bank, testing was the poor relation in project budgets, and it was an uphill battle to get people onboard.
At the retail chain, I did a spreadsheet that could do ad-hoc reports directly from the Mercury testing database, and the test script steps numbers ranged from a couple up to 140+, when they should have a mean of 10-12 with a narrow standard deviation, so testing practice is still relatively immature in places.
Of course a software house will run up their stuff on a new OS preview to see where the land lies, but these later OSs have been going through several changes in functionality at the 11th hour, unlike previous ones, where the candidate release was fairly final more than a month before the official release to OEMs. With that sort of short timeframe, it doesn’t leave much time to fix anything that may be broken, especially if it requires a fix from the OS maker, as it did for Win 10 for Cubase to get the go ahead.
To a certain extent, it is a crap-shoot as to whether one makes it through without a holdup, as even if one’s program is ‘perfect’, it only takes something to change incorrectly on the OS side at the final release, and it may adversely affect yours, but the competition gets the Ok. Again, it all comes down to how its done under the hood in yours. And unless there is a complete rewrite from the ground up for each OS version, there is a tonne of legacy code that has managed to dodge bullets for years! And Cubase is the most full featured DAW out there, so has more possible failure vectors.