phases: avoid N² behavior in `advanceboundary`
We allowed duplicated entries in the deque, which each entry could potentially
insert all its ancestors. So advancing boundary for the full repository would
mean each revision would walk all its ancestors, resulting in O(N²) iteration.
For repository of any decent size, N² is quickly insane.
We introduce a simple set to avoid this and get back to reasonable performance.
#require svn13
$ cat <<EOF >> $HGRCPATH
> [extensions]
> mq =
> [diff]
> nodates = 1
> [subrepos]
> allowed = true
> svn:allowed = true
> EOF
fn to create new repository, and cd into it
$ mkrepo() {
> hg init $1
> cd $1
> hg qinit
> }
handle svn subrepos safely
$ svnadmin create svn-repo-2499
$ SVNREPOPATH=`pwd`/svn-repo-2499/project
$ SVNREPOURL="`"$PYTHON" $TESTDIR/svnurlof.py \"$SVNREPOPATH\"`"
$ mkdir -p svn-project-2499/trunk
$ svn import -qm 'init project' svn-project-2499 "$SVNREPOURL"
qnew on repo w/svn subrepo
$ mkrepo repo-2499-svn-subrepo
$ svn co "$SVNREPOURL"/trunk sub
Checked out revision 1.
$ echo 'sub = [svn]sub' >> .hgsub
$ hg add .hgsub
$ hg status -S -X '**/format'
A .hgsub
$ hg qnew -m0 0.diff
$ cd sub
$ echo foo > a
$ svn add a
A a
$ svn st
A* a (glob)
$ cd ..
$ hg status -S # doesn't show status for svn subrepos (yet)
$ hg qnew -m1 1.diff
abort: uncommitted changes in subrepository "sub"
[255]
$ cd ..