- Added support for `-F configfile'. This enables using a specially
crafted muttrc for automated testing.
- Centralized call to _get_cword
- Used _get_pword
- Specified non-wordbreak characters to _get_cword() & _get_pword() for
bash-4.
- Added automated tests for -f and -A
- Fixed test suite's assert_complete() (test/library.exp) to drop only
the last element
This fixes a bug under bash-4 where completing:
scp -F 'spaced conf' <TAB>
causes `dequote' to yield errors:
bash: eval: line 1: unexpected EOF while looking for matching `''
bash: eval: line 2: syntax error: unexpected end of file
The bug occurs because of a bug in bash-4.0, where quoted words are split
unintended, see: http://www.mail-archive.com/bug-bash@gnu.org/msg06095.html
Workaround is now to silence `dequote' in case of errors and wait for bash-4 to
be fixed...
Changed `out' parameter of `assert_bash_exec()' to accept also -1 and 0:
@param mixed $out (optional) Reference to variable to hold output.
If variable equals -1 (default) the bash command is
expected to return no output. If variable equals 0,
any output from the bash command is disregarded.
This fixes the situation of commit cfcf9fa where contrib/ri was causing
invisible errors when running the test suite.
- Added code comments to _get_cword, __get_cword3 & __get_cword4
- (testsuite) Added tests for _get_cword
- (testsuite) Bugfixes assert_bash_exec() && match_items()
Bash-4 splits COMP_WORDS using characters from COMP_WORDBREAKS, but has
a bug where quoted words are also splitted, see:
http://www.mail-archive.com/bug-bash@gnu.org/msg06095.html
__get_cword3 is used for bash-2/3 and __get_cword4 is used for bash-4.
__get_cword4 handles well temporarily disabling of COMP_WORDBREAK
characters, but fails quoted words (a 'b c) and subshells (a $(b c).
See the expected failures when running the automated tests.
__get_cword3 does a better job of returning quoted words.
To run the automated tests on bash-3/4:
$ ./runUnit _get_cword.exp [--tool_exec <path to bash-3/4 binary>]
Using option `--tool_exec' you can now specify which bash binary you
want to use:
$ ./runUnit --tool_exec /opt/bash-4/bin/bash
$ ./runCompletion --tool_exec /opt/bash-4/bin/bash
If not specified, current `bash' as found in the tcl path (/bin/bash)
will be used.
assert_env_unmodified() now calls `diff_env' with `LAST-ARG' as the last
argument. Now $_ becomes `LAST_ARG' and $_ doesn't interfere with the
diff anymore as it would if $_ was the sed script.
Added line continuation (\) to bash script of `get_hosts_avahi()'.
Bash < 3.2.41 has a bug where 'history' disappears from SHELLOPTS
whenever a shopt setting is sourced or eval'ed. Disabling 'history'
makes it not show in tests "Environment should not be modified"
for bash < 3.2.41.
Added helper function `is_bash_version_minimal()' to
test/lib/library.sh.
- Fix get_hosts_avahi() if avahi-browse not available
- Removed test "Tab should complete hostnames" from finger since it's
returning (for the moment) undetectable ip-addresses from known_hosts
files. Furthermore the next test "Tab should complete partial
hostname" is testing the same.
Quote sed output ("\1") retrieving "GlobalKnownHostsFile/UserKnownHostsFile" to
prevent bash globbing of special characters, e.g. '*'.
Added unit test "Config file containing star (*) should work".
Added test library function `assert_bash_list()'.
To run unit tests:
cd test && ./runUnit _known_hosts_real.exp
This mimics the old behaviour where you could reuse `_known_hosts()' as
a helper function and pass it `-a' or `-c'.
NOTE: Using `_known_hosts' as a helper function and passing options to
`_known_hosts' is deprecated: Use `_known_hosts_real' instead.
Fixed `assert_complete()' to remove an optional (partial) argument from
`cmd' when expecting `cmd + longestCommonPrefix'.
Added `prompt' argument to procs `assert_bash_exec()' &
`sync_after_int()'
The local timeout settings `-timeout 1' in some loops seemed to cause
inpredictable results, e.g. with `./runCompletion dkpg.exp' since this
test has a long list to match. Sometimes the `spawn_id' was gone before
the `teardown' method was called.
Removing the short timeout seems to give more predictable results.
Fix `assert_complete' to expect the longest common prefix to be presented as completion.
When trying to do a:
package require textutil::string
it appeared DejaGnu overwrites `unknown', the tcl function that is called to attempt autoloading. A workaround is used for tcl auto-loading to work: `init.tcl' is reloaded to restore the original `unknown' function:
source [file join [info library] init.tcl]
See also: http://fvue.nl/wiki/DejaGnu:_proc_"tclPkgUnknown_..."_does_not_exist
`match_items()' now uses bash sort instead of tcl lsort.
The difference in sorting happens because bash and tcl use a different sorting order:
* bash sort uses the current locale's collating sequence and character set
* tcl lsort uses a C-like sort order where each character is sorted according to its numeric value
See also: http://fvue.nl/wiki/Tcl:_Lsort_output_differs_from_bash_sort
This fix allows the `dpkg -L' test to compare all packages. To run the test:
./runCompletion dpkg.exp
- First argument shouldn't complete with commands
- Tab should complete partial hostname
To run the tests: $ cd test && ./runCompletion ssh.exp
Added test library function: assert_complete_partial
