Bitrot -- loss of LFS

dyson at iquest.net dyson at iquest.net
Thu Nov 27 08:57:01 PST 2003


Answering Terry's comment about LFS:

I had about 90% of the LFS development complete
(rewritten to eliminate much of the unnecessary
and inefficient copying.)  At that time, Kirk
had started softupdates, but I also KNEW and
UNDERSTOOD the limitations of LFS.

In essense, after CAREFULLY reading and understanding
Ganger and Patt (and knowing about Kirk's work),
I had decided that LFS would have been a folly.
(It would have been a maintenance issue, people would
have 'gotten used' to the feature, and would have
been a wasted allocation of development resources.)

LFS (as existent in 4.4) was an interesting experiment,
and was excluded (by me) because of the amount of
ongoing support (including bringing it up to product
quality.)

If you look at the original code -- you'll notice
ALOT of unnecessary copying and primitive memory
management.  It was DEFINITELY a good research
project, and perhaps worthwhile if softupdates
hadn't happened.

John



More information about the freebsd-current mailing list