This fixes the issue with nonexistant directories (some Linux
distributions put these for local modules), and also fixes the
issue of dot meaning any character instead of simply dot.
# The first commit's message is:
Simplify default fish_prompt
No need for the set_color caching now that it's a builtin.
Also simplify the 3 classic prompts in fish_config's sample_prompts set.
Remove comment that AFAICT is not true anymore.
Ensure someone setting __fish_active_key_bindings as a universal
variable doesn't screw up the initial keybinding load.
All opam subcommands and descriptions are covered, along with
all the flags that are common to all commands. However, only
`opam config` has complete subsubcommand coverage.
Apparently, in zsh, Meta+H can be used to display the manpage for
the current command. This commit adds this zsh feature to fish shell.
The F1 keybinding is left, although it's now secondary according to
fish help, as some terminal emulators don't let the user press F1 key.
Use the new `read -z` flag to complete git aliases better. This approach
won't break if an alias contains a newline.
Also fix stash completion, which was broken on BSD sed.
Setting a non-existant path component to PATH logs an error to stderr.
This is not appropriate for non-interactive temporary modifications,
like the one done by the `sudo` completion helper function.
In the base config.fish, fish_function_path and fish_complete_path have
$__fish_datadir/{functions,completions} added to them if not already
present. For some reason they were replacing the final path component
instead of being added on to the end.
The new --wraps functionality was breaking aliases of the form
`alias foo='bar baz'`. That is, aliases where the body is multiple
words. Extract the first word of the body and use that instead.
Use better errors for aliases with no name or no body.
Add the --wraps option to 'complete' and 'function'. This allows a
command to (recursively) inherit the completions of a wrapped command.
Fixes#393.
When evaluating a completion, we inspect the entire "wrap chain" for a
command, i.e. we follow the sequence of wrapping until we either hit a
loop (which we silently ignore) or the end of the chain. We then
evaluate completions as if the wrapping command were substituted with
the wrapped command. Currently this only works for commands, i.e.
'complete --command gco --wraps git\ checkout' won't work (that would
seem to encroaching on abbreviations anyways). It might be useful to
show an error message for that case.
The commandline builtin reflects the commandline with the wrapped
command substituted in, so e.g. git completions (which inspect the
command line) will just work. This sort of command line munging is
also performed by 'complete -C' so it's not totally without precedent.
'alias will also now mark its generated function as wrapping the
'target.
- Require all requests to use a session path.
- Use a redirect file to avoid exposing the '/start' URL on the
command line, as it contains the cookie value.
Fix for CVE-2014-2914.
Closes#1438.
Currently fish doesn't recognize toor as special. However, it's likely
that on BSD systems, fish shell will be used on toor, not on root (toor
is an intentionally existing account to use more advanced shell on, like
shell).
Use `functions -q` instead of searching the `functiosn -na` list for the
provided word. This may result in an automatically-loaded function being
sourced, but that happens anyway with the default output.
This change means the results of `test -q foo` can be relied upon to
indicate whether `foo` can actually be invoked. Previosly, if `foo` was
the name of an automatically-loaded function file but did not actually
define a function `foo`, and there was no execuable `foo`, then `type -q
foo` would lie and say `foo` can be invoked when it can't.
The --quiet flag is useful when only the exit status matters.
Fix the documentation for the -t flag to no longer claim that `type` can
print "keyword", as it never does that.
Stop printing a blank line for functions/builtins when the -p flag has
been passed. It's just not useful.
Track whether -a and -f have been supplied separately. That way both
`type -a -f command` and `type -f -a command` behaves correctly, as does
`type -a -f foo` where there are multiple executables named `foo` in the
$PATH.
Stop using getopt to parse flags. It's far more expensive than
necessary, and results in long flags not being parsed on OS X. This also
allows args starting with - after the options list to be properly
interpreted as a value to test.
Print the error message to stderr as is appropriate.
Use the new `command -p` functionality when the -a flag has not been
provided (`command` does not have any equivalent to the -a flag),
instead of using `which`. This is faster and also avoids any possible
disagreement between `which` and what fish thinks is valid.
Stop testing every path to see if it's executable, that test has already
been done by `which` or `command -p`.
The end result is `type -P ls` is roughly 250% faster, according to
profiling, on my OS X machine.
* use $XDG_CACHE_HOME for __fish_print_packages completion caches
* when starting fishd, redirect fishd output to /dev/null, not a
predictable path
Fix for CVE-2014-3219.
Closes#1440.
When you chroot in Debian, bash shows the chroot environment in the prompt:
```bash
...
if [ -z "${debian_chroot:-}" ] && [ -r /etc/debian_chroot ]; then
debian_chroot=$(cat /etc/debian_chroot)
fi
PS1='${debian_chroot:+($debian_chroot)}\u@\h:\w\$ '
...
```
This is the effect:
```
(chroot_env) user@host:~#
```
It is useful when chrooting, since usually the hostname remains the same and thus you can't distinguish where you are.
Fix for CVE-2014-2906.
Closes a race condition in funced which would allow execution of
arbitrary code; closes a race condition in psub which would allow
alternation of the data stream.
Note that `psub -f` does not work (#1040); a fix should be committed
separately for ease of maintenance.
Closes#1437
The if statement checking the output of hg bookmarks uses two conditions
joined by the or keyword. However, only the first part was being used.
Wrapping the two statements with begin and end properly combines them.
At some point the non-verbose, non-informative variant of the prompt
(e.g. the variant that looks like the bash prompt) was modified to try
and show the behind/ahead counts the same way the informative prompt
does. Besides being wrong, it also didn't work because behind/ahead
weren't defined.
