view tests/test-status-rev.t @ 23401:fd1bab28a8cc stable

manifest: fix a bug where working copy file 'add' mark was buggy

Because the same dictionary was used to (1) get node from parent and (2) store
annotated version, we could end up with buggy values. For example with a chain
of renames:

  $ hg mv b c
  $ hg mv a b

The value from 'b' would be updated as "<old-a>a", then the value of c would be
updated as "<old-b>a'. With the current dictionary sharing this ends up with:

    '<new-c>' == '<old-a>aa'

This value is double-wrong as we should use '<old-b>' and a single 'a'.

We now use a read-only value for lookup. The 'test-rename.t' test is impacted
because such a chained added file is suddenly detected as such.

Tests of 'hg status --rev <rev>' to make sure status between <rev> and '.' get
combined correctly with the dirstate status.

Sets up a history for a number of files where the filename describes the file's
history. The first two letters of the filename describe the first two commits;
the third letter describes the dirstate for the file. For example, a file called
'amr' was added in the first commit, modified in the second and then removed in
the dirstate.

These codes are used for commits:
x: does not exist
a: added
c: clean
m: modified
r: removed

These codes are used for dirstate:
d: in dirstate, but deleted from disk
f: removed from dirstate, but file exists (forgotten)
r: removed from dirstate and disk
q: added, but deleted from disk (q for q-rious?)
u: not in dirstate, but file exists (unknown)

  $ hg init
  $ touch .hgignore
  $ hg add .hgignore
  $ hg commit -m initial

First letter: first commit

  $ echo a >acc
  $ echo a >acd
  $ echo a >acf
  $ echo a >acm
  $ echo a >acr
  $ echo a >amc
  $ echo a >amd
  $ echo a >amf
  $ echo a >amm
  $ echo a >amr
  $ echo a >ara
  $ echo a >arq
  $ echo a >aru
  $ hg commit -Aqm first

Second letter: second commit

  $ echo b >xad
  $ echo b >xaf
  $ echo b >xam
  $ echo b >xar
  $ echo b >amc
  $ echo b >amd
  $ echo b >amf
  $ echo b >amm
  $ echo b >amr
  $ hg rm ara
  $ hg rm arq
  $ hg rm aru
  $ hg commit -Aqm second

Third letter: dirstate

  $ echo c >acm
  $ echo c >amm
  $ echo c >xam
  $ echo c >ara && hg add ara
  $ echo c >arq && hg add arq && rm arq
  $ echo c >aru
  $ hg rm amr
  $ hg rm acr
  $ hg rm xar
  $ rm acd
  $ rm amd
  $ rm xad
  $ hg forget acf
  $ hg forget amf
  $ hg forget xaf
  $ touch xxu

Status compared to one revision back

  $ hg status -A --rev 1 acc
  C acc
BROKEN: file appears twice; should be '!'
  $ hg status -A --rev 1 acd
  ! acd
  C acd
  $ hg status -A --rev 1 acf
  R acf
  $ hg status -A --rev 1 acm
  M acm
  $ hg status -A --rev 1 acr
  R acr
  $ hg status -A --rev 1 amc
  M amc
BROKEN: file appears twice; should be '!'
  $ hg status -A --rev 1 amd
  ! amd
  C amd
  $ hg status -A --rev 1 amf
  R amf
  $ hg status -A --rev 1 amm
  M amm
  $ hg status -A --rev 1 amr
  R amr
  $ hg status -A --rev 1 ara
  M ara
BROKEN: file appears twice; should be '!'
  $ hg status -A --rev 1 arq
  R arq
  ! arq
  $ hg status -A --rev 1 aru
  R aru
  $ hg status -A --rev 1 xad
  ! xad
  $ hg status -A --rev 1 xaf
  $ hg status -A --rev 1 xam
  A xam
  $ hg status -A --rev 1 xar
  $ hg status -A --rev 1 xxu
  ? xxu

Status compared to two revisions back

  $ hg status -A --rev 0 acc
  A acc
  $ hg status -A --rev 0 acd
  ! acd
BROKEN: file exists, so should be listed (as '?')
  $ hg status -A --rev 0 acf
  $ hg status -A --rev 0 acm
  A acm
  $ hg status -A --rev 0 acr
  $ hg status -A --rev 0 amc
  A amc
  $ hg status -A --rev 0 amd
  ! amd
BROKEN: file exists, so should be listed (as '?')
  $ hg status -A --rev 0 amf
  $ hg status -A --rev 0 amm
  A amm
  $ hg status -A --rev 0 amr
  $ hg status -A --rev 0 ara
  A ara
  $ hg status -A --rev 0 arq
  ! arq
  $ hg status -A --rev 0 aru
  ? aru
  $ hg status -A --rev 0 xad
  ! xad
BROKEN: file exists, so should be listed (as '?')
  $ hg status -A --rev 0 xaf
  $ hg status -A --rev 0 xam
  A xam
  $ hg status -A --rev 0 xar