I have found some stuff the I could use as is. Single functions, app notes, libraries. A big plus over the days before the net where if the local book store did not cover the subject you had to hammer at it till it worked.
I have never found anything that just worked. It always had to be adjusted to get it going. But I i think I was ahead at the end. I also had some stuff I thought was garbage and ended up redoing it. And other stuff I looked at and just said no.
It is like garage sale shoppers. there is gold out there, but you have to sift through it.

But if I need it that bad I will buy it. If I pay I expect it to work ( yes have been disappointed). In the 1-wire case I could look and see if the chip worked, and what it said. Then I used it to validate the work. Look the Maxium's reader matches mine. So I am good (yes I scoped it too).
No yes or no answer here for me. Bad working code for some odd protocol may save more time then reading a bad spec. Even undocumented tools may help ( look I am off a bit, oooh MSB first).

The value of free varies.

Poor or non working app notes and eval boards still amuse me. Not really free but come on they want you to buy.