1.8 suboptimal rep-cache deduplication
From: Philip Martin <philip_at_codematters.co.uk>
Date: Thu, 08 Mar 2018 20:14:08 +0000
Philip Martin <philip_at_codematters.co.uk> writes:
> My reproduction doesn't trigger the bug in 1.8 but that seems to be
Comparing 1.8 and 1.9 the reason that 1.8 doesn't do deduplication is
/* We don't want 0-length PLAIN representations to replace non-0-length
In 1.9 the corresponding code is transaction.c:get_shared_rep:2234
if ( ((*old_rep)->expanded_size != rep->expanded_size)
Since 1.8 doesn't have the expanded_size check it sets *old_rep to NULL
Index: subversion/libsvn_fs_fs/fs_fs.c
the deduplication happens (and the checksum failure does not occur
-- PhilipReceived on 2018-03-08 21:14:15 CET |
This is an archived mail posted to the Subversion Dev mailing list.
This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.