> When you compare Linux and Windows. Are you comparing Linux server you support and Windows desktop you support? Conclusion cannot be draw from such settings because server usage and desktop usage have totally different patterns.
You are 100% right.
> Have you supported Linux desktop also? In what scale?
Yes, but, yes, in much smaller scale (1% = 30 desktops, largely of the same hardware types as the Windows desktops), but with some exclusivity (I am part of a team of 30 people that gives suppport to the Windows desktops, but I am the sole responsible for the support to the Linux desktops, and the Mac support team is me plus one colleague).
> Since you seems to be quite experienced with driver issues, Cool. Still have some questions.
> From my experience, the initial setup of Windows can be erratic. But once a working setup is there, it's very unlikely to have driver issues again. Do you think so? In what situation can you get a setup with driver issues from a non-savvy user?
It is my experience that, in an enterprise, non-savvy users do not install operating systems. We do that in the lab, for the new batch of computers, get it right, and clone the others.
> For Linux, well, despite the lack of formal support (I know servers have formal support, but desktops and laptops don't), kernel and/or Xorg upgrade can have random effects. I guess it is pretty safe stay with RHEL/CentOS, though desktop application installation can be quite hard there. But if it is Arch, even Ubuntu, things can be "fun".
You don't upgrade critical Windows major versions without review (hell, we don't even let users' machines install the Windows shutdown-time hotfixes without reviewing them). If you keep the same procedure for Linux, you'll be safe too.
> In fact, some pirated distribution of Windows (the most famous one may be Tomato Garden) get automatic driver installation working at least 99% of times many years ago. People seldom bother with driver issues afterwards. So it is a bit surprised to me that you seems to suffering from it still.
I believe you (seriously). More ahead.
> Even if genuine Windows may require manual driver installation, Dell/HP should have listed nicely working drivers for models on their website.
Dell and HP are actually the culprits, driver-wise speaking. They change their models' chipsets mid-batch and their drivers are not always perfectly aligned. These are the source of many of our headaches. Sometimes some other hardware's drivers (I'm looking at you, Lexmark) causes evil drivers interactions, too.
> For the reasons 1-5 you mentioned. I guess developer experience is a key factor.
Yes, it is. Availability, too. (more ahead, again :D)
> From a surface level, you may ask your fellows about their opinions on Lazarus. I look forward to hearing your answer.
I use Lazarus here, the initial fellow-developer reaction is "that's Delphi, isn't it?" :D Now, seriously, we don't migrate everyone to Lazarus because my city has a score of Delphi consultants that we can contract if something goes wrong or if we have a spike on development needs...
That and politics. Recently, thanks to some database-driver problems, we had to migrate the database-middleware-layer away from BDE. To Zeos, you ask? No, to "pay-for-upgrades, pay-for-every-bug-correction" UniDAC. Go figure.