I just loaded my new subversion database successfully with the
dumpfile from my new PinHandler branch. And you don't belive it: the
final comparison of roughly 20600 files (~3GB of data) didn't show a
single bit of difference.
Did I mention, that you need to have enough memory installed on the
machine while running the conversion ;-) Esp. if you have large binary
files, that are all committed into the same revisions. Since the code
keeps all these records in memory until the revision is finished, the
process can slow down tremendiously if the os starts to swap. I'm
currently running with a "working set size" of 1GByte and I still have
3000 revisions to go. It also seems, that the memory is not really
returned to the OS, probably due to heap corruption.
Since the price of memory isn't that high anymore, this isn't really a
problem. Perhaps I should try on our win64 machine with 8GB physical
memory ;-)
Dirk
_______________________________________________
vss2svn-users mailing list
Project homepage:
http://www.pumacode.org/projects/vss2svn/
Subscribe/Unsubscribe/Admin:
http://lists.pumacode.org/mailman/listinfo/vss2svn-users-lists.pumacode.org
