> > > and does anyone want a challenge of porting tdb to apr?
> > > *grin*
> >
> > Challenge, did somebody say challenge? I'm always up for a challenge.
> > :-)
:) i just checked the codebase. it uses:
lseek
fcntl
read
write
open
close
mmap (if supported)
munmap (if supported)
malloc (3 times)
free (16 times)
realloc (once, in tdb_expand(), in TDB_INTERNAL mode. you don't _have_
to use a file with tdb, you can do in-memory only).
so, not exactly a _big_ challenge, which is the kind of challenges
i like: small and gratifying and part of something larger.
the larger:
cool code that could provide fast, SHARED in-memory, multi-reader
multi-writer simultaneous access, and oh yeah, it also has the
option to do files as well as in-memory.
the keys, however, are a unique (void*, size_t) tuple,
and so is the data.
the bad news:
does anyone want to contact the three remaining authors,
rusty russell, andrew tridgell ([EMAIL PROTECTED]) and
jeremey allison ([EMAIL PROTECTED]) to see if they'd
like to release a version under the APR license?
also, anton blanchard (sparc32 kernel hacker Extreme :)
who wrote the spinlock code (no, it's not on the
list of copyrights, i used to work at same company as
him when he was writing it).
whoever decides to contact them, please DO NOT mention
my name, it will have the OPPOSITE intended effect, i
guarantee that, 100%.
luke
