[Gnu-arch-users] Re: arch lkml

[Miles Bader]

address@hidden (Eric W. Biederman) writes:
> > One thing _is_ clear: the current representation is a _huge_ win if
> > something goes wrong, because it doesn't use a big-binary-blob (or multiple
> > smaller-binary-blobs) every storage, everything is in clear common formats.
> 
> Hmm.  gzip tar files are binary blobs, just in a common format.

Perhaps my point wasn't clear.

Arch's (current) storage format:

 (1) Uses the filesystem in a natural way, with each changeset (a fairly
     small unit) in a separate file, divided into directories organized
     along branches/categories/etc.

 (2) Puts changesets into .tar.gz files, with a completely straight-
     forward and obvious organization (even to someone that doesn't know
     anything about arch).

 (3) Essentially never, ever, touches old files, it just creates new ones.

Some of the advantages of this representation are:

 * You can _easily_ (without installing _any_ new tools, and _without_
   documentation) look over the whole archive, see how it's organized
   and get all your data out.  If all the executables, sources,
   documentation, and hackers of all arch implementations ever written
   suddenly exploded, you could _still_ get your data, and do something
   useful with it.  A good hacker with superficial knowledge of arch
   could probably even reverse-engineer a new arch implementation just
   from looking at an archive.

 * Because of (3), a bug in arch is unlikely to ever trash your archive;
   a bug in the filesystem code could do it, but that's _much_ less
   likely [and of course will affect any other representation too].

 * As with CVS, if you _really_ want to, you can go in and _edit_
   arch archives, to do something arch doesn't support, and you can do
   this with normal unix shell scripts and tools.  Clearly this is not
   a very safe thing to do, but in the real world, sometimes this sort
   of thing is necessary, and with arch, it's at least a little bit
   less dangerous.

The `binary blob' I was contrasting with is where the repository is
stored in a single big file, with a non-standard binary-format.  This
sort of representation is _much_ scarier.

-Miles
-- 
If you can't beat them, arrange to have them beaten.  [George Carlin]