8 This finds all scripts in the test directory named test-* and executes

     6 See http://www.selenic.com/mercurial/wiki/index.cgi/WritingTests for

     9 them. The scripts can be either shell scripts or Python. Each test is

     7 more information on writing tests.

    10 run in a temporary directory that is removed when the test is complete.

    11 

    12 A test-<x> succeeds if the script returns success and its output

    13 matches test-<x>.out. If the new output doesn't match, it is stored in

    14 test-<x>.err.

    15 

    16 There are some tricky points here that you should be aware of when

    17 writing tests:

    18 

    19 - hg commit and hg merge want user interaction

    20 

    21   for commit use -m "text"

    22   for hg merge, set HGMERGE to something noninteractive (like true or merge)

    23 

    24 - changeset hashes will change based on user and date which make

    25   things like hg history output change

    26 

    27   use commit -m "test" -u test -d "1000000 0"

    28 

    29 - diff and export may show the current time

    30 

    31   use -D/--nodates to strip the dates

    32 

    33 - You can append your own hgrc settings to the file that the environment

    34   variable HGRCPATH points to. This file is cleared before running a test.

    35 

    36 You also need to be careful that the tests are portable from one platform

    37 to another.  You're probably working on Linux, where the GNU toolchain has

    38 more (or different) functionality than on MacOS, *BSD, Solaris, AIX, etc.

    39 While testing on all platforms is the only sure-fire way to make sure that

    40 you've written portable code, here's a list of problems that have been

    41 found and fixed in the tests.  Another, more comprehensive list may be

    42 found in the GNU Autoconf manual, online here:

    43 

    44     http://www.gnu.org/software/autoconf/manual/html_node/Portable-Shell.html

    45 

    46 sh:

    47 

    48 The Bourne shell is a very basic shell.  /bin/sh on Linux is typically

    49 bash, which even in Bourne-shell mode has many features that Bourne shells

    50 on other Unix systems don't have (and even on Linux /bin/sh isn't

    51 guaranteed to be bash).  You'll need to be careful about constructs that

    52 seem ubiquitous, but are actually not available in the least common

    53 denominator.  While using another shell (ksh, bash explicitly, posix shell,

    54 etc.) explicitly may seem like another option, these may not exist in a

    55 portable location, and so are generally probably not a good idea.  You may

    56 find that rewriting the test in python will be easier.

    57 

    58 - don't use pushd/popd; save the output of "pwd" and use "cd" in place of

    59   the pushd, and cd back to the saved pwd instead of popd.

    60 

    61 - don't use math expressions like let, (( ... )), or $(( ... )); use "expr"

    62   instead.

    63 

    64 grep:

    65 

    66 - don't use the -q option; redirect stdout to /dev/null instead.

    67 

    68 - don't use extended regular expressions with grep; use egrep instead, and

    69   don't escape any regex operators.

    70 

    71 sed:

    72 

    73 - make sure that the beginning-of-line matcher ("^") is at the very

    74   beginning of the expression -- it may not be supported inside parens.

    75 

    76 echo:

    77 

    78 - echo may interpret "\n" and print a newline; use printf instead if you

    79   want a literal "\n" (backslash + n).

    80 

    81 false:

    82 

    83 - false is guaranteed only to return a non-zero value; you cannot depend on

    84   it being 1.  On Solaris in particular, /bin/false returns 255.  Rewrite

    85   your test to not depend on a particular return value, or create a

    86   temporary "false" executable, and call that instead.

    87 

    88 diff:

    89 

    90 - don't use the -N option.  There's no particularly good workaround short

    91   of writing a reasonably complicated replacement script, but substituting

    92   gdiff for diff if you can't rewrite the test not to need -N will probably

    93   do.