garbage collection - Understanding git gc --auto -

i'm experimenting aggressive auto gc in git, packing purposes. in repos if git config --list have setup

... gc.auto=250 gc.autopacklimit=30 ... 

if git count-objects -v get

count: 376 size: 1251 in-pack: 2776 packs: 1 size-pack: 2697 prune-packable: 0 garbage: 0 

but git gc --auto doesn't change these figures, nothing being packed! shouldn't loose objects packed since i'm 126 objects on gc.auto limit?

one of main points of gc --auto should quick, other commands can call “just in case”. achieve that, object count guessed. git config says under gc.auto:

when there approximately more many loose objects in repository […]

looking @ code (too_many_loose_objects() in buildin/gc.c), here’s happens:

1. the gc.auto divided 256 , rounded up
2. the folder contains objects start 17 opened
3. it checked if folder contains more objects result of step 1

this works fine, since sha-1 evenly distributed, “all objects start x” representative whole set. of course works big big amount of objects. lazy maths, guess @ least >3000. 6700 (the default value of gc.auto), should work quite reliably.

the core question me why need such low setting , whether important runs @ 250 objects. setting of 250, gc run have 2 loose objects start 17. chance happens > 80% 600 objects , > 90% 800 objects.

update: couldn’t – had math :). wondering how estimation system work. here’s plot of results. given gc.auto, how high probability gc start when there gc.auto (red) / gc.auto * 1.1 (green) / gc.auto * 1.2 (orange) / gc.auto * 1.5 (blue) / gc.auto * 2 (purple) loose objects in repo?