changegroup: avoid large copies
- handle chunk headers separately rather than prepending them to
(potentially large) chunks
- break large chunks into 1M pieces for compression
- don't prepend file metadata onto (potentially large) file data
#!/bin/sh
echo "[extensions]" >> $HGRCPATH
echo "hgext.convert=" >> $HGRCPATH
hg init orig
cd orig
echo foo > foo
echo bar > bar
hg ci -qAm 'add foo and bar' -d '0 0'
hg rm foo
hg ci -m 'remove foo' -d '0 0'
mkdir foo
echo file > foo/file
hg ci -qAm 'add foo/file' -d '0 0'
hg tag -d '0 0' some-tag
hg log
cd ..
hg convert orig new 2>&1 | grep -v 'subversion python bindings could not be loaded'
cd new
hg out ../orig
echo '% dirstate should be empty:'
hg debugstate
hg parents -q
hg up -C
hg copy bar baz
echo '% put something in the dirstate:'
hg debugstate > debugstate
grep baz debugstate
echo '% add a new revision in the original repo'
cd ../orig
echo baz > baz
hg ci -qAm 'add baz'
cd ..
hg convert orig new 2>&1 | grep -v 'subversion python bindings could not be loaded'
cd new
hg out ../orig
echo '% dirstate should be the same (no output below):'
hg debugstate > new-debugstate
diff debugstate new-debugstate
echo '% no copies'
hg up -C
hg debugrename baz