[darcs-users] hashed-storage index

[Petr Rockai]

Eric Kow <kowey at darcs.net> writes:

> I just meant if reading the large index file would contribute a
> significant amount (say a few seconds) of time to our total wall time in
> a large repository (as you say, it's still a net improvement, but I'm
> just curious).  I was wondering if the future would require us to break
> the index up into smaller pieces.  But from the sounds of it, we won't
> need to be worrying about that, at least not right now.
Actually, there's no difference between reading one seven-megabyte file or
seven one-megabyte files.

> One more question: where do we get the times that we store in the index
> from?  My guess is that we get it from the working copy, because getting
> it from the pristine would just bring us back to the original problem?
Working copy -- these are the only meaningful timestamps around, anyway. The
index keeps hashes of the working files and uses the timestamp/filesize combo
to keep those up to date. These are then used for fast hash-based subtree/file
comparisons.

Yours,
   Petr.

-- 
Peter Rockai | me()mornfall!net | prockai()redhat!com
 http://blog.mornfall.net | http://web.mornfall.net

"In My Egotistical Opinion, most people's C programs should be
 indented six feet downward and covered with dirt."
     -- Blair P. Houghton on the subject of C program indentation