Perens: There are times when Linus Torvalds can be a real idiot, and this is one of these times
More fallout from the whole BitKeeper affair at the Reg:
With Andrew Tridgell silent, apparently on legal advice, open source community leader Bruce Perens has stepped up to defend the work Tridgell did reverse engineering the protocols used by Bitkeeper. Bitkeeper is the closed source proprietary source code management tool that until last week, Linus Torvalds used to manage Linux kernel source code.
Torvalds has responded to the controversy by blaming Tridgell, and made two extraordinarily intemperate attacks on him online, the second accusing him of willful destruction.
Perens told us that Torvalds needs to "cool it."
"There are times when Linus Torvalds can be a real idiot, and this is one of these times," said Perens.
He pointed out that the closed source tool was foisted on kernel developers despite the consensus that it was inappropriate for a GPL project. Many declined to use it regardless of the gateways McVoy created, says Perens. So the criticism is "very severely unfair" to Tridgell, he says.
As for McVoy, who Perens has known since he was seven (the pair grew up on Long Island), the advice is emphatic.
"Larry could have left gracefully, but I'm afraid that's just not in him... If he has any sense, he'll shut up."
Sorry, I still think Linus is right here. The fact is, closed source, open source, whatever, BitKeeper was chosen because it was the best tool for the job. Linus wanted something that could manage many disperate small patches, didn't rely on a base repository and could work in an entirely distributed P2P fashion. That is still BK. This whole holy war that somehow the Open Source effort has to remain idealogically pure is just retarded.
Comments
RE: Perens: There are times when Linus Torvalds can be a real i
it was the best tool for the job providing you paid for the full version (or were given a full version as Linus was).
Without the full version (i.e. with the free(beer) version the linux developers were using) then you had no access to metadata on the files, making it much harder to use it for source management. (e.g. regression testing was difficult).
So what were the developers supposed to do? All pay for the full version (which *was* the best tool for the job), but who has that sort of money?
Or, alternatively, they could try to back out the metadata from BK, which is what tridge was trying to do.
RE: Perens: There are times when Linus Torvalds can be a real i
Well, I understand that it was pricey -- much too pricey, really. I think the single seat was $5,500. Really, though, I imagine at OSDL, IBM, RedHat etc that isn't a big deal.
I am really curious how many of the mainline kernel developers DIDN'T have a full copy paid for by their employers. I know that there is a KISS army of driver developers and such, but I would wager that at the top echelon maybe 3-5% of people don't work somewhere that bought them a copy anyway.
RE: Perens: There are times when Linus Torvalds can be a real i
I would also note, however, that even the free version I would wager is better than doing it by email, which would be the once and future method.