It 's fixed. My test corpus was an rsync'ed directory containing all the RFCs and Internet Drafts from ref-editor.org.
It takes about 20 minutes to do the initial checkin, on a FreeBSD machine (1GB RAM, 2GHz Celeron) Thanks! Ge' D. Richard Hipp wrote: > On Feb 25, 2009, at 3:42 PM, Gé Weijers wrote: > > >> On a related note: I have a large collection of files (13000), some of >> which are large PDFs, most are source code etc. I have been unable to >> check this tree in in fossil, it runs out of memory (RAM) after a few >> minutes, after the working set size goes above half a gigabyte. I can >> add the files just fine, but the final commit is painful. >> > > > http://www.fossil-scm.org/fossil/ci/f6790b7c3cc237fac3d0cc064cb262d50fa12265 > > I tested with the Enron email corpus. That's over half a million > separate files totaling 1.5 GB in size. Took about an hour, but the > commit did go through (after the fix above), and memory usage never > went above 50MB or so. > > I'll continue to do performance testing on this enron repository and > see if I can't get commits to run a little faster. Bear in mind, > though, that each fossil check-in stores an MD5 checksum over every > file in the check-in (both those that changes and those that have not) > and regardless of what you do, it does take some time to run MD5 over > 1.5 GB in 517431 separate files. > > > D. Richard Hipp > [email protected] > > > > _______________________________________________ > fossil-users mailing list > [email protected] > http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users > _______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
