Re: [PATCH] fetch-pack: do not mix --pack_header and packfile uri

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Jonathan Tan <jonathantanmy@xxxxxxxxxx> writes:

> One simplification that we could do is to eliminate the unpack-objects
> codepath. As far as I understand, the main advantage of writing loose
> objects is that we have automatic SHA-1 collision detection, but we have
> such mitigations when writing packs too, so that might not be as large a
> benefit as we think. This simplification would have enabled us to avoid
> this bug, I think.

My understanding is that the primary advantage of loose objects
codepath is to help us avoid having too many little packs (instead,
we can accumulate enough objects in the loose form and let GC pack
them, at least the ones among them that are still reachable, into a
single pack).  Historically, the only mode of operation "repack"
offers that reduces the number of remaining packs has been "do full
reachability of the entire history, and pack everything into one",
so avoiding creation of little packs and leaving things loose until
we accumulate enough used to matter.

With the geometric rolling repacking, it may not matter as much, and
keeping everything packed, even in a small pack, might start to be
overall win.  So I am not opposed to such a simplification; we may
not be ready for it right now, but I think it would be a sensible
future direction.







[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]

  Powered by Linux