How to control which directories get statistics?

Ben Escoto
Mon, 27 May 2002 11:40:43 -0700

Content-Type: text/plain; charset=us-ascii

Before each directory would get its own directory_statistics file.
These currently look something like this:

SourceFiles 2066
SourceFileSize 29500712 (28.1 MB)
MirrorFiles 1
MirrorFileSize 4096 (4.00 KB)
NewFiles 2065
NewFileSize 29496616 (28.1 MB)
DeletedFiles 0
DeletedFileSize 0 (0 bytes)
ChangedFiles 1
ChangedSourceSize 4096 (4.00 KB)
ChangedMirrorSize 4096 (4.00 KB)
IncrementFileSize 0 (0 bytes)

However, I just realized that these can take up a lot of space.  For
instance, I seem to have about 13000 directories on my system.  Since
each directory_statistics file probably takes up 4096 or 8192 bytes
(whatever the minimum block size is), these statistics files are
taking up 50-100MB per session.  Nominally this is several times the
amount of space taken up by my actual increment files.

    So it seems the default should be not to create all these
statistics files.  However, sometimes per directory statistics might
be useful.  I see 3 possibilities so far:

1.  All or nothing switch --save-directory-statistics

2.  Option "--directory-statistics-levels n" which means to save
    statistics for directories at most n levels below the root.

3.  Option --directory-statistics-regexp, so statistics are saved for
    the directories matching the given regexp.

I'm leaning towards (3) even though it is the most complicated (and
there are some non-obvious things, for instance that the mirror's path
should match the regexp, not the source path).  Opinions?

Ben Escoto

Content-Type: application/pgp-signature

Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Exmh version 2.5 01/15/2001