Re: 'svnadmin recover' with FSFS revprop packing: edge case bug

2010-02-01 Thread Daniel Shahaf
I've filed this as issue #3571: http://subversion.tigris.org/issues/show_bug.cgi?id=3571 Patches welcome... Daniel Daniel Shahaf wrote on Sun, 31 Jan 2010 at 00:00 +0200: > fs_fs.c:recover_body() needs to be taught that the youngest revision > does not necessarily have a revprop file. The foll

Re: 'svnadmin recover' with FSFS revprop packing: edge case bug

2010-02-01 Thread Daniel Shahaf
Julian Foad wrote on Mon, 1 Feb 2010 at 11:54 -: > On Sun, 2010-01-31, Daniel Shahaf wrote: > > fs_fs.c:recover_body() needs to be taught that the youngest revision > > does not necessarily have a revprop file. ... > > Found by: svnadmin_tests 13, when built with > > -DSVN_FS_FS_DEFAULT_MAX_FIL

Re: 'svnadmin recover' with FSFS revprop packing: edge case bug

2010-02-01 Thread Julian Foad
On Sun, 2010-01-31, Daniel Shahaf wrote: > fs_fs.c:recover_body() needs to be taught that the youngest revision > does not necessarily have a revprop file. The following repository > demonstrates the bug: > > % cat db/format > 5 > layout sharded 4 > % cat db/current > 3 >

'svnadmin recover' with FSFS revprop packing: edge case bug

2010-01-30 Thread Daniel Shahaf
fs_fs.c:recover_body() needs to be taught that the youngest revision does not necessarily have a revprop file. The following repository demonstrates the bug: % cat db/format 5 layout sharded 4 % cat db/current 3 % find db/rev* db/revprops db/revprops/revprops.db