Re: [Gnu-arch-users] Re: Working out a branching scheme [was: tag --seal --fix]

[Tom Lord]

    > From: James Blackwell <address@hidden>

    >> Having 10'000 revisions on a single branch is something that people
    >> want to do with CVS, and will want to do with arch.  If arch is to be
    >> useful for those people, it should be able to deal gracefully with
    >> that sort of thing.

    > > If you think those people are wrong in wanting to do that, your answer
    > > should hopefully be ``that's insane, but arch will handle it just 
fine.''

    > Arch "works fine" with thousands of revisions in one branch. In fact,
    > there are already a small handful of archives that already do this. 

    > For example, Robert Collins has converted the entire cvs history of
    > automake into arch, which has thousands of revisions. 

    > Arch handles the large number of revisions gracefully, gradually slowing
    > down as more and more revisions are added.


At the same time, were autoconf _developed_ under arch the archive
wouldn't come out that way and it's a weakness of the cscvs tools that
they don't know how to cope with this issue yet.


(Thanks for the fun stats.)


-t






    > > (Another thing which I find arch unsuitable for is hacking at a large
    > > remote tree when you don't have much local disk space.  I spent a few
    > > years working on a 300 MB tree with a laptop with 4 GB local disk,
    > > typically none of which was free; that would not have been possible
    > > with arch.)
    > 
    > This problem is already in the past and as time continues to march on,
    > the problem will fade even further. Today, 300 megabytes is a large 
archive;
    > seven years ago it was *huge*.
    > 
    > These are facts:
    > 
    >    1. archives grow linearly, with a shallow slope.
    >    2. drive space grows nonlinerarly, doubling every 2 years or so.
    >    3. Arch is ready for all but the very largest archives *today*
    >    4. The average archive size today is 16MB.
    >    5. The average revision size is about 50K (the median is 17K) 
    >    6. The average archive has 11 versions
    > 
    > We can probably agree on this:
    > 
    >    7. A project is 'very popular' if it averages 220 revisions a day 
    >       (or 20 revisions for each of its 11 versions)
    > 
    >                                Today    Two years        Four Years
    > Archive size                   16MB     23.3MB (1)       30.6MB (2)
    > Drive Space for a new machine   80GB     160GB            320GB
    > Ratio of Drive space used       0.0195%  0.0142%          0.0093%
    > 
    > (1) 16   + 2 * (20*11*17/1024)
    > (2) 23.3 + 2 * (20*11*17/1024)
    > 
    > Even if you map it out to your extraordinarly large archive, we get:
    > 
    >                                Today    Two years        Four Years
    > Archive size                   300MB     307.3MB (3)     314.6MB (4)
    > Drive Space for a new machine   80GB     160GB            320GB
    > Ratio of Drive space used       0.366%   0.188%          0.096%
    > 
    > (3) 300   + 2 * (20*11*17/1024)
    > (4) 307.3 + 2 * (20*11*17/1024)
    > 
    > As you can see, there's no problem 
    > 
    > -- 
    > James Blackwell          Please do not send me carbon copies of mailing
    > Smile more!              list posts. Such mail is unsolicited. Thank you!
    > 
    > GnuPG (ID 06357400) AAE4 8C76 58DA 5902 761D  247A 8A55 DA73 0635 7400
    > 
    > 
    > _______________________________________________
    > Gnu-arch-users mailing list
    > address@hidden
    > http://mail.gnu.org/mailman/listinfo/gnu-arch-users
    > 
    > GNU arch home page:
    > http://savannah.gnu.org/projects/gnu-arch/
    > 
    >