Edmund Horner wrote:
> Øyvind A. Holm wrote:
> > But now it works again. I shut down the httpd server while installing
> > new svn versions, maybe that has something to say? The checkins was done
> > via file:// , I'll restart the script to see if the same thing happens
> > again. It could also be a problem with just passing the magical 65536th
> > revision directly.
>
> My speculation (unsupported by much knowledge or any real evidence) is
> that it's to do with skip deltas, and the amount of work needed to do
> *whatever* on reaching the 65536th revision: many of the previous
> revisions need to be redone to point to r65536.
>
> In particular, from looking at the arrow diagram in
> http://pasky.ji.cz/~pasky/dev/pavs/tree/delta-theory.txt, r1 depends on
> *16* other revisions; perhaps some part of the deltafication mechanism
> has never been tested with that many revisions at once.
>
> I committed a small change (alternating between "0" and "1") to the same
> file, 65536 times, as in Øyvind's recipe, and it went pretty smoothly (I
> cheated by using svnadmin load for most of it :-) until the 65536th
> revision, which took several minutes to complete and did a lot of
> writing to c:\pagefile.sys (which sounded a bit odd to me). The
> operation eventually completed successfully.
>
> PS. Is there an official location of ghudson's document cited above? It
> didn't look like a Subversion or ghudson URI ;-)
I believe the cannonical location for that document is
http://web.mit.edu/ghudson/thoughts/file-versioning
-garrett
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri May 28 19:22:55 2004