On Mon, 27 Sep 2010, Jan KrÃger wrote:
> In 479b56ba ('make "repack -f" imply "pack-objects --no-reuse-object"'),
> git repack -f was changed to include recompressing all objects on the
> zlib level on the assumption that if the user wants to spend that much
> time already, some more time won't hurt (and recompressing is useful if
> the user changed the zlib compression level).
>
> However, "some more time" can be quite long with very big repositories,
> so some users are going to appreciate being able to choose. Hence, this
> adds a new -F option that uses the old behaviour of recalculating deltas
> only and keeping the zlib compression intact.
>
> Measurements taken using this patch on a current clone of git.git
> indicate a 17% decrease in time being made available to users:
>
> git repack -Adf 38.79s user 0.56s system 133% cpu 29.394 total
> git repack -AdF 34.84s user 0.56s system 145% cpu 24.388 total
>
> Signed-off-by: Jan KrÃger <jk@xxxxx>
> ---
>
> The concrete case that prompted me to write this patch was a repository
> of 25 GB that some guys were trying to repack. 17% of the time needed to
> repack -f that much data is... substantial.
>
> Discussion point: it might make more sense to switch the meanings
> around, making -F do the 'bigger' routine and reverting -f to what it
> used to be. I don't feel strongly about that, however.
That's exactly what I was about to propose before I read through your
email down to this part.
I personally don't find --no-reuse-object particularly useful. I hardly
imagine that people are changing the pack compression level that often
if at all. So I doubt moving the current --no-reuse-object behavior to
-F and reverting -f to --no-reuse-delta would cause any serious
inconvenience. It certainly won't _break_ anything. So you have my ACK
to do that change.
In addition to that change, perhaps a note could be added to the
documentation for pack.compression indicating that for the new setting
to take effect for existing packs, they must be repacked with -F.
Nicolas