Thread (4 messages) flat view 4 messages, 2 authors, 2016-06-15

git-gc "--aggressive" somewhat broken

From: Linus Torvalds <torvalds@linux-foundation.org>
Date: 2016-06-15 22:43:20

This flag seems misdesigned as-is.

It makes the window bigger (good, and aggressive), but it also enabled 
"-f".

Which sometimes causes _worse_ packing.

In particular, there's a gcc import at

	git://git.infradead.org/toolchain/gcc.git

which apparently packs down quite nicely to a 450MB pack, but when it gets 
repacked "aggressively", it blows up to 1.4GB. The reason? That pack was 
generated using --depth=100 --window=100 (I think).

So under certain circumstances, "--aggressive" is anything _but_ 
aggressive, and actually causes much worse packing.

If we want to be really aggressive, we migth decide to pass a new flag to 
pack-objects that does something closer to what "aggressive" was meant to 
do: it would use existing delta's if they exist, but _despite_ existing it 
could look if there are even better choices.

So right now we have:

 - default behaviour:

   always re-use existing deltas, don't look at alternatives at all.

   This is optimal for CPU/memory/IO usage, and is generally a good idea

 - "-f" (and as a result, the current bad "git gc --aggressive"):

   never re-use existing deltas, always look for new ones.

   This is good if you have reason to believe that the old choices are 
   bad, or you need to force re-generation of deltas (because you want to 
   force a new pack-file format, for example)

and the missing piece might be

 - "git pack-objects --aggressive":

   re-use existing deltas _and_ look for even better ones.

   This is good if all you're looking for is better packing.

Hmm?

		Linus
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help