sqliterevmap: break ".hashes()" cycle in a safer way The `fromsvn()` revset implementation could cause weakref error when using sqliterevmap like: File "hgsubversion/util.py", line 357, in <lambda> return subset.filter(lambda r: tonode(r) in hashes) File "hgsubversion/maps.py", line 542, in __contains__ return self.get(key) != None File "hgsubversion/maps.py", line 533, in get for row in self.revmap._query( ReferenceError: weakly-referenced object no longer exists Basically the seemingly harmless assignment could break surprisingly: # dangerous: `hashes` does not have a reference of `meta.revmap` and may # become unavailable after `meta`, `revmap` being released by refcount. hashes = meta.revmap.hashes() The above syntax is nice to support while avoiding cycles is also nice. This patch removes `revmap._hashes` so the revmap no longer owns a reference of a `ReverseRevMap` object so the `ReverseRevMap` object no longer needs to use weakref for `self.revmap`. This could actually be caught by `comprehensive/test_sqlite_revmap.py`. I was not careful enough verifying the "fromsvn()" patch.

#!/bin/sh
#
# Generate binaryfiles.svndump
#

mkdir temp
cd temp

mkdir project-orig
cd project-orig
mkdir trunk
cd ..

svnadmin create testrepo
svnurl=file://`pwd`/testrepo
svn import project-orig $svnurl -m "init project"

svn co $svnurl project
cd project/trunk
# Add a regular binary file, and an unflagged one
python -c "file('binary1', 'wb').write('a\0\0\nb\0b')"
python -c "file('binary2', 'wb').write('b\0\0\nc\0d')"
svn add binary1 binary2
svn propset svn:mime-type application/octet-stream binary1
svn propdel svn:mime-type binary2
svn ci -m 'add binaries'
# Update them
python -c "file('binary1', 'wb').write('a\0\0\nc\0d')"
python -c "file('binary2', 'wb').write('b\0\0\0\nd\0e')"
svn ci -m 'change binaries'
# Remove them
svn rm binary1 binary2
svn ci -m 'remove binaries'
cd ../..

svnadmin dump testrepo > ../binaryfiles.svndump