I don't read a lot of modern patents, for the obvious reasons: there are legal risks, and the patent texts are useless even if the invention isn't. But there used to be a time when a software patent wasn't necessarily an obfuscated hodgepodge of legalese with no apparent technical content.
A system description doesn't get much better than that. If somebody couldn't implement a GC based on that patent, they're not a competent systems programmer. (It was also definitely novel. Non-obviousness is harder to argue 25 years after the fact).
Why on earth would there be legal risks to reading a patent? Contrary to copyright, whether you're infringing on a patent because you read it or because you happened to have the same idea is completely irrelevant. I know some free software guys are pretty paranoid about reading proprietary code, but I see no reason why this should transpose to patents.
(but IANAL, so perhaps I'm missing some subtle point)
https://en.wikipedia.org/wiki/Patent_infringement_under_Unit... 'If an infringer is found to have deliberately infringed a patent (i.e. "willful" infringement), then punitive damages can be assessed up to three times the actual damages. Legal fees can also be assessed.'
Right. It might be just a matter of legal departments cargo culting, but it's very common for companies to ask engineers not to read outside patents for this reason. One might think that it'd make sense to read through patents in order to avoid them, but that only works if the patents are actually even remotely understandable by a normal person.
There are situations where patents need to be read, but it's not going to be done by a random engineer.
Weak argument for obviousness: As I understand it, the basic idea is to use a copying GC that pins everything pointed to by stack or registers. I had the same idea earlier this week, after spending a few days' spare cycles thinking about GC and having read virtually none of the literature.
Consider for example the (expired, so should be safe to read) mostly-copying GC patent: http://www.google.com/patents/US4907151
A system description doesn't get much better than that. If somebody couldn't implement a GC based on that patent, they're not a competent systems programmer. (It was also definitely novel. Non-obviousness is harder to argue 25 years after the fact).