Tue, 16 Sep 2008
Kernel Summit, Day 1
The first day of the Kernel Summit was pretty interesting, altough against my expectations it was much more development process orineted than technically oriented.
Probably the most interesting was Linus' remark to the discusssion about storage, especially SSD: he said something like "Forget the erase block sizes and whatnot, and treat the SSD just as a really fast HDD. It is not optimal for the current hardware, but it will be for the upcoming one".
I wonder what has changed in the last year or so, because from previous discussions on this topic it seemed that HW manufacturers were even willing to remove even the wear-levelling controllers from SSDs, provided that the OS could handle wear levelling on its own (which Linux does, via the MTD stack). But I guess there are still plenty of use cases which for compatibility reasons essentially require the FAT filesystem (cameras, etc.), and that in turn require at least the wear levelling in HW. So probably the development will go exactly the opposite direction than was anticipated a year ago.
Another interesting presentation was Rafal Wysocki's statistics on regression discovery and fix times, from which he even predicted the optimal release time of the next stable version.
Spent the evening with Evgeniy Polyakov and Herbert Xu discussing things from Russian politics in Georgia through motor boats to the X11 architecture shortcomings. Ah, BTW: Linus' car has an interesting registration number :-)