[Discuss] Possible backup scenarios for a 500GB drive

Alan W. Irwin irwin at beluga.phys.uvic.ca
Tue Oct 30 10:52:08 PDT 2007


On 2007-10-30 08:46-0700 Adam Parkin wrote:

> Honestly though, I have *never* had any problems with CVS and thus I've 
> always been a bit puzzled at why people knock it so much.

I have had a number of "episodes" with cvs over the years.  The worst was just
after joining the PLplot project in 2000 and right in the middle of learning
cvs, I discovered a bunch of what I called "zombie" files in the repository
that kept rising from the dead (or descending from the Attic in this case).
If you did a cvs checkout, no problems.  But when doing any branch
development these files automatically became part of any new branch.  Also,
they arose from the dead when you tried cvs export.  Somehow, earlier
versions of the cvs software had put the repository (started in the early
90's) into an inconsistent state.  By doing a careful compare of the cvs
export tree with the cvs checkout tree and deleting the zombie files for a
second time (even though they were already in the Attic).  Of course, this
story may only be a side note about historical cvs code quality since one
could claim the error was probably in the mid-90's and fixed since, but I
don't know for sure.  However, I do know there was at least one long-term
bug for cvs with the more modern SourceForge version.  Very occasionally,
you would make a commit, and the lockfile would not be removed leaving
nobody with access to the file until the SF staff intervened by hand.  Because
it was such an intermittent error it was extremely difficult to track down,
but I assume the SF staff finally nailed it because these problems quit in
the last year or so.

Also, have you really looked at all the cvs options and how their meaning
changes depending on position in the command line?  Clearly a sign of old
crufty code that desperately needed a rewrite.  That rewrite was the
separate subversion project (since there was no community taking
responsibility for the cvs code, just individual deperate programmers
looking for a way to fix bugs such as the lockfile problem).

If you already like cvs, I predict you will be delighted by svn just like I
was after putting up for years with the cvs quirks.  Read the appendix of
the svn book on the cvs/subversion differences.  Do the quick start in that
book and actually use svn for a single project.  You won't want to go back
to cvs for very long. (That's what happened to me, I joined a SF project
which was already using svn.  After a day or two I realized there was no way
I was going back to cvs so I converted my remaining SF projects from cvs to
svn. PLplot was the biggest of those conversions, but I managed to do it in
a way that still preserved all the history of those many cvs commits since
the early 90's which completely satisfied the originators of PLplot who are
still members of the PLplot development team.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of
Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________


More information about the Discuss mailing list