Re: [Gnu-arch-users] Why we might use subversion instead of arch.

[Brian May]

>>>>> "Robert" == Robert Collins <address@hidden> writes:

    >> Arch lets me cache if I understand correctly. My objection is
    >> more philosophical:
    >> 
    >> The build source should match some set of files on the
    >> repository.
    >> 
    >> (pretend you don't trust patch. In that case, you'd really like
    >> a snapshot of HEAD to always be stored somewhere...)

    Robert> Thats bogus. Neither CVS nor Bitkeeper, nor RCS store a
    Robert> plaintext copy of the leading edge of each branch. AFAIK
    Robert> Arch is the only one that allows you to store a full,
    Robert> plaintext copy of a revision. (Thats what a cached
    Robert> revision is).

The big difference with arch vs. CVS/bitkeeper/subversion, is that
arch exposes more of the internals to the user. With the others, the
internals are kept hidden from the user, and in fact some are kept
hidden from the client via a communications protocol. (sidenote: soon
arch might also get its own communications protocol too).

You could argue this isn't as good as far as UI goes. There is a mind
set that we need to make computer programs as simple as possible to
use for the untrained user, and this would appear to go against that
trend. This is a can of worms, I will stop now.

I think the arch approach leads to greater flexibility. It means I can
write my own scripts that do things I want instead of being forced to
comply with a published interface. I also find it possible to debug
errors, and get at least a good guess what happened. With subversion,
I have on rare occasions got weird errors and never understood why;
instead I was forced to delete the checked out tree and start again.
-- 
Brian May <address@hidden>