Philip Martin wrote on Thu, Mar 28, 2013 at 12:09:30 +0000: > Stefan Sperling <s...@elego.de> writes: > > > If we decide that verifying transactions before commit is a good > > thing, I'd much rather see a new 'svnlook verify' command that can > > be run from the pre-commit hook, instead of this implementation. > > A separate process could be a performance hit since it would not share > the FSFS cache used by the commit process. On the other hand a separate > process might be an advantage if there is a caching bug.
That's good. There is already a comment in svn_fs_fs__verify_root() asking how to disable/bypass caching; "use a child process" is a good start... Do we have any per-machine caches? Or only per-process and memcached?
