> I can only imagine in that era there was quite a bit more learning forced upon you than these days with our plethora of (somewhat) standards-compliant hardware and in-tree drivers.
It's romanticized a lot, but boy was it painful! hear my story and weep.
In 1995 I got slackware with a textbook from the university library. I had a brand-new 486 paid for by a dad with 2 other kids in college and nothing more than a factory job to pay for things.
So my 486 had no CDROM. And it only had 4MB of RAM. And a pirated version of DOS, until I got slackware.
After 3 months of saving from weekend work I made enough to get a shitty soundblaster pro clone. And then I returned it for another unit, because, as I complained to the store owner "this doesn't work". When the replacement didn't work either I realised it must be my computer, more specifically, those "drivers" provided on the floppy disk were for msdos :-/
So then I started reading a bit, and digging into the driver sources, and reading the little pamphlet that came with the soundcard. "Hah!" I thought, "the IRQ channel for the soundcard appears to be hardwired!"
And thus begun the longest 3-day weekend of my short life up to that point; Linux had no modules, everything was compiled into a single binary image. You have any idea how long it takes a lowly 486 with 4MB of RAM to compile a kernel?
The entire weekend was "make this change, type make $SOMETHING, then wait for 50m, then watch it fail to boot, then examine how far the bootloader got, then boot with previous image, then undo that change and repeat the process".
Eventually had the soundcard working though. On a system on which no bloody games would run ...