Re: [Gnu-arch-users] On incompetent uses of English

[Robert Collins]

On Fri, 2005-12-09 at 10:13 -0800, Derek Zhou wrote:
> Robert Collins <address@hidden> writes:
> 
> > In arch 1 'resnapping' the inode signature is definately doable, if you
> > care to. In the simplest form just store the {hash of choice} in the
> > index file for the revision, and consult that to determine if content
> > has changed. Oh and store the permissions for the files and reapply
> > those. Oh, and there are a couple more little gotchas but nothing
> > insurmountable. 
> 
> Resnapping is certainly doable, but not easy. To make resnapping safe,
> you need to hash _everything_, not just file contents, but also
> permissions, directories (including empty ones), and symbolic
> links. AFAIK arch 1 is in maintainance mode, and this look like not
> a trivial thing to do and may introduce many bugs. The idea of my
> patch is if it is slow anyway, at least we can automate the
> slowness. I have no objection to move the stale revlib entries out of
> the way but not "rm -rf" it though.    

There is an index file in revision library entries that is amenable to
recording this. The on disk changes are very modest and easy to
complete. (For compatability consider a new parallel index file).

The internal changes if this is kept to a clear scope are also modest
and straight forward. I would say 4-5 days to test it, write it &
(optionally) write an upgrade script to upgrade existing revlib entries.

In terms of bugs, that would depend where and how you do it. Moving an
active revlib entry is really spectacularly unsafe given how often it
should occur - have you considered alternatives like marking it as
unusable and skipping over it to prior revlib entries? And when you find
one make a pristine from that and build forward?

Still, I'm really not interested in debating this with you, I've offered
my opinion, HTH HAND.

Rob

signature.asc
Description: This is a digitally signed message part