Using int was a mistake. This patch changes String, StringImpl,
StringView and StringBuilder to use size_t instead of int for lengths.
Obviously a lot of code needs to change as a result of this.
If we can't find an executable to exec() after forking, we don't want
to run the atexit() handlers in the child process. Just use _exit()
instead to avoid this.
This was causing us to write out the shell history to ~/.history every
time a "command not found" error was printed.
This patch reduces the O(n) tab completion to something like O(log(n)).
The cache is just a sorted vector of strings and we binary search it to
get a string matching our input, and then check the surrounding strings
to see if we need to remove any characters. Also we no longer stat each
file every time.
Also added an #include in BinarySearch since it was using size_t. Oops.
If `export` is called, we recache. Need to implement the `hash` builtin
for when an executable has been added to a directory in PATH.
Previously, we did not properly restore termios settings
after running built-in commands. This has been fixed
by ensuring that we only change the termios settings
when we are forking a child process.
Shell.cpp uses its own line discipline which handles
echoing and line editing. Because of this we disable
ICANON and ECHO so that we don't get duplicate characters
or weird line editing errors.
We also revert these settings just before running a command.
This is so that commands may run with proper line editing
and echoing.
Added a few builtin functions to the shell to make navigating a bit
easier in the terminal.
`pushd` allows a user to "push" the current directory to the directory
stack, and then `cd` to the new directory.
`popd` allows the used to take the directory on the top of the stack
off before `cd`'ing to it.
`dirs` gives the state of the current directory stack.
This is only a partial implementation of the `bash` version
(gnu.org/software/bash/manual/html_node/Directory-Stack-Builtins.html)
, and doesn't include any of the +N or -N commands as of yet.
This patch allows passing a script as an argument to the Shell program.
We will read the specified line by line and pass them through the Shell
command interpreter.
This is not very powerful, but it's a start :^)
Make LineEditor::get_line() responsible for printing the prompt. That way
we can re-prompt after clearing the screen on ^L.
This makes the Serenity Terminal feel a little bit more like home :^)
This makes waitpid() return when a child process is stopped via a signal.
Use this in Shell to catch stopped children and return control to the
command line. :^)
Fixes#298.
When resizing the terminal, we now clear the entire current line and reset
the shell's LineEditor input state. This makes it look and feel kinda the
same as xterm.
Fixes#286.
A glob has to be resolved against the directory corresponding to
the part of the path it is found in, not the current directory.
For example, in /usr/i*/AK/, the glob has to be resolved inside
/usr. Moreover, an argument can contain more than one glob, such
as /u*/*/?, in which case they have to be resolved recursively.
In case a glob matches nothing, the argument should be used as is.
After reading a bunch of POSIX specs, I've learned that a file descriptor
is the number that refers to a file description, not the description itself.
So this patch renames FileDescriptor to FileDescription, and Process now has
FileDescription* file_description(int fd).
This was unnecessarily confusing. When we build up a chain of commands
connected by pipes, we now store the file descriptors of each end of
these pipes as rewirings in a vector. The rewirings are then put into
effect by calls to dup2().
If I'm understanding the standard C library correctly, setenv() copies while
putenv() does not. That's really confusing and putenv() basically sucks.
To know which environment variables to free on replacement and which ones to
leave alone, we keep track of the ones malloced by setenv in a side table.
This patch also moves Shell to using setenv() instead of putenv().
Fixes#29.
This is in keeping with how putenv should function. It does mean that
the shell's export command now leaks, but that's not a difficult fix.
Contributes to #29.
