[Gnu-arch-users] Re: ANN: Linux Kernel Arch Repository

[John Goerzen]

Karel Gardas <address@hidden> writes:

> Huh! This sounds like _big_ task. Could you be so kind and provide us
> with some numbers like:

Sure.

> - whole archive size

497MB spanning 2383 files or directories.  202MB is consumed by cached
revs, of which there are 11, meaning actual changesets consume 295MB
of space.  763 patches are represented in the archive.  The most
recent revision contains 15,850 files and directories exclusive of
arch control data.

Recent kernel revs typically change 300-1000 files per revision.  Some
go closer to 1300 to 1500.  (These are numbers from a brief sampling)

Of other interest: I downloaded a tar.bz2 from kernel.org for
approximately one in 20 kernel revs, and used diff.bz2's to import all
the others.  (the tar.bz2 was as sort of a sanity check to make sure
nothing weird got missed by a diff.bz2).  This had no impact on how
things were stored in arch, but it may be of interest that my
collection of these files was 1.2GB.  Coincidentally, I had attempted
this same project in Subversion a few months back and its archive was
about 1.1GB.  (That one never went live publically because of hosting
problems)

> - time needed to do whole import

It took me about a week of work evenings / weekends.  Probably more
time was spent figuring out exactly what the developers did when they
renamed things than actual tla processing.  However, once you get to
about 2.3.x, tla starts to get pretty slow with the commits.

> - revision library size for whole archive - if you even do have it

I do not keep a revision library for this project :-)  

-- John