# HG changeset patch
# User Mads Kiilerich <madski@unity3d.com>
# Date 1494281490 -7200
# Node ID 78fb569e2c339742cd6066958adc55bca593be63
# Parent  a0e46f6b248b7a6db0f3cd09b1dba6dcefeb8fd4
graft: test coverage of grafts and how merges can break duplicate detection

This demonstrates unfortunate behaviour: extending the graft range cause the
graft to behave differently. When the graft range includes a merge, we fail to
detect duplicates that are ancestors of the merge.

diff -r a0e46f6b248b -r 78fb569e2c33 tests/test-graft.t
--- a/tests/test-graft.t	Mon May 08 23:05:01 2017 -0400
+++ b/tests/test-graft.t	Tue May 09 00:11:30 2017 +0200
@@ -1309,4 +1309,49 @@
   $ hg status --change .
   M b/x
 
+Prepare for test of skipped changesets and how merges can influence it:
+
+  $ hg merge -q -r 1 --tool :local
+  $ hg ci -m m
+  $ echo xx >> b/x
+  $ hg ci -m xx
+
+  $ hg log -G -T '{rev} {desc|firstline}'
+  @  7 xx
+  |
+  o    6 m
+  |\
+  | o  5 y
+  | |
+  +---o  4 y
+  | |
+  | o  3 x
+  | |
+  | o  2 b
+  | |
+  o |  1 x
+  |/
+  o  0 a
+  
+Grafting of plain changes correctly detects that 3 and 5 should be skipped:
+
+  $ hg up -qCr 4
+  $ hg graft --tool :local -r 2::5
+  skipping already grafted revision 3:ca093ca2f1d9 (was grafted from 1:13ec5badbf2a)
+  skipping already grafted revision 5:43e9eb70dab0 (was grafted from 4:6c9a1289e5f1)
+  grafting 2:42127f193bcd "b"
+
+Extending the graft range to include a merge will unfortunately make us miss
+that 3 and 5 should be skipped:
+
+  $ hg up -qCr 4
+  $ hg graft --tool :local -r 2::7
+  skipping ungraftable merge revision 6
+  skipping already grafted revision 5:43e9eb70dab0 (was grafted from 4:6c9a1289e5f1)
+  grafting 2:42127f193bcd "b"
+  grafting 3:ca093ca2f1d9 "x"
+  note: graft of 3:ca093ca2f1d9 created no changes to commit
+  grafting 7:d3c3f2b38ecc "xx"
+  note: graft of 7:d3c3f2b38ecc created no changes to commit
+
   $ cd ..