tests/test-convert-cvsnt-mergepoints

findrenames: Optimise "addremove -s100" by matching files by their SHA1 hashes.

We speed up 'findrenames' for the usecase when a user specifies they
want a similarity of 100% by matching files by their exact SHA1 hash
value. This reduces the number of comparisons required to find exact
matches from O(n^2) to O(n).

While it would be nice if we could just use mercurial's pre-calculated
SHA1 hash for existing files, this hash includes the file's ancestor
information making it unsuitable for our purposes. Instead, we calculate
the hash of old content from scratch.

The following benchmarks were taken on the current head of crew:

addremove 100% similarity:
  rm -rf *; hg up -C; mv tests tests.new
  hg --time addremove -s100 --dry-run

  before:  real 176.350 secs (user 128.890+0.000 sys 47.430+0.000)
  after:   real   2.130 secs (user   1.890+0.000 sys  0.240+0.000)

addremove 75% similarity:
  rm -rf *; hg up -C; mv tests tests.new; \
      for i in tests.new/*; do echo x >> $i; done
  hg --time addremove -s75  --dry-run

  before: real 264.560 secs (user 215.130+0.000 sys 49.410+0.000)
  after:  real 218.710 secs (user 172.790+0.000 sys 45.870+0.000)

#!/bin/sh

"$TESTDIR/hghave" cvs || exit 80

filterpath()
{
    eval "$@" | sed "s:$CVSROOT:*REPO*:g"
}

cvscall()
{
    echo cvs -f "$@"
    cvs -f "$@"
}

# output of 'cvs ci' varies unpredictably, so discard most of it
# -- just keep the part that matters
cvsci()
{
    echo cvs -f ci -f "$@"
    cvs -f ci -f "$@" 2>&1 | egrep "^(new|initial) revision:"
}

hgcat()
{
    hg --cwd src-hg cat -r tip "$1"
}

echo "[extensions]" >> $HGRCPATH
echo "convert = " >> $HGRCPATH
echo "graphlog = " >> $HGRCPATH

echo "% create cvs repository"
mkdir cvsmaster
cd cvsmaster
CVSROOT=`pwd`
export CVSROOT
CVS_OPTIONS=-f
export CVS_OPTIONS
cd ..
filterpath cvscall -Q -d "$CVSROOT" init

echo "% checkout #1: add foo.txt"
cvscall -Q checkout -d cvsworktmp .
cd cvsworktmp
mkdir foo
cvscall -Q add foo
cd foo
echo foo > foo.txt
cvscall -Q add foo.txt 
cvsci -m "add foo.txt" foo.txt
 
cd ../..
rm -rf cvsworktmp

echo "% checkout #2: create MYBRANCH1 and modify foo.txt on it"
cvscall -Q checkout -d cvswork foo

cd cvswork

cvscall -q rtag -b -R MYBRANCH1 foo
cvscall -Q update -P -r MYBRANCH1
echo bar > foo.txt
cvsci -m "bar" foo.txt
echo baz > foo.txt
cvsci -m "baz" foo.txt

echo "% create MYBRANCH1_2 and modify foo.txt some more"
cvscall -q rtag -b -R -r MYBRANCH1 MYBRANCH1_2 foo
cvscall -Q update -P -r MYBRANCH1_2

echo bazzie > foo.txt
cvsci -m "bazzie" foo.txt

echo "% create MYBRANCH1_1 and modify foo.txt yet again"
cvscall -q rtag -b -R MYBRANCH1_1 foo
cvscall -Q update -P -r MYBRANCH1_1

echo quux > foo.txt
cvsci -m "quux" foo.txt

echo "% merge MYBRANCH1 to MYBRANCH1_1"
filterpath cvscall -Q update -P -jMYBRANCH1
# carefully placed sleep to dodge cvs bug (optimization?) where it
# sometimes ignores a "commit" command if it comes too fast (the -f
# option in cvsci seems to work for all the other commits in this
# script)
sleep 1
echo xyzzy > foo.txt
cvsci -m "merge1+clobber" foo.txt

echo "% return to trunk and merge MYBRANCH1_2"
cvscall -Q update -P -A
filterpath cvscall -Q update -P -jMYBRANCH1_2
cvsci -m "merge2" foo.txt

REALCVS=`which cvs`
echo "for x in \$*; do if [ \"\$x\" = \"rlog\" ]; then echo \"RCS file: $CVSROOT/foo/foo.txt,v\"; cat $TESTDIR/test-convert-cvsnt-mergepoints.rlog; exit 0; fi; done; $REALCVS \$*" > ../cvs
chmod +x ../cvs
PATH=..:${PATH} hg debugcvsps --parents foo | sed -e 's/Author:.*/Author:/' -e 's/Date:.*/Date:/'

cd ..