Daniel Shahaf wrote: > Julian Foad wrote on Fri, Apr 27, 2012 at 22:04:45 +0100: >> One point being the semantics is different: _bind_revnum sets column to >> NULL for INVALID_REVNUM. >> >> Don't know if INVALID_REVNUM ever can hit these two particular code >> paths. > > You could check that by throwing an assert() there and running the test > suite?
Could do; haven't done. Greg said on IRC 'no reason for them, other than "code written before revnum binding code"', so I just changed them to _revnum in r1331733. - Julian >> > Index: subversion/libsvn_wc/entries.c >> > =================================================================== >> > + /* ### Shouldn't this use __bind_revnum? (If not, say why.) */ >> > SVN_ERR(svn_sqlite__bind_int64(stmt, 7, node->revision)); >> > Index: subversion/libsvn_wc/wc_db.c >> > =================================================================== >> > + /* ### Shouldn't this use __bind_revnum? (If not, say why.) */ >> > SVN_ERR(svn_sqlite__bind_int64(stmt, 7, piwb->original_revnum));
