For one of my projects I am using Makefile to carry out some tasks. But for some reason the system is not capturing my input or I might doing something wrong. Here is my code:
#read -e -p "Please enter email-addresses: " -i "user1#domain.com,user2#domain.com" EMAIL_ADDRESSES
#echo $EMAIL_ADDRESSES;
#echo $$EMAIL_ADDRESSES;
#echo ${EMAIL_ADDRESSES};
#echo $${EMAIL_ADDRESSES};
But here is my output:
MAIL_ADDRESSES
[Blank]
[Blank]
[Blank]
What am I doing wrong? How do I fix this?
First, make always run its recipes in /bin/sh, not /bin/bash. Some of the capabilities you're using here for read are specific to Bash and are not available in standard POSIX shells.
Second, make runs every individual line in a recipe in a different shell. So any shell variables you set in one line are lost when the shell exits and are not available in the next line. If you want to preserve them you need to put the entire script on a single (logical) line, like this:
#printf 'Please enter email-addresses: '; \
read EMAIL_ADDRESSES; \
echo $$EMAIL_ADDRESSES
If you really want to use Bash features you should probably invoke it directly.
Lastly, it's generally a bad idea to use standard input from a makefile. If someone ever wanted to run your makefile with parallel jobs enabled then all but one command will not have any standard input (it will be redirected from /dev/null). It's better to ask the user to provide the value on the command line as a make variable assignment.
Related
Using the code below, I am able to redirect the program's standard error output to an error.log file, and call a specific function based on whether or not the program exits peacefully. However in addition to this, I would like to organize the stderr output using the fold command, to keep the lines under a max length of 80 characters. How can I achieve this, whilst retaining the current functionality?
Code:
define func
./myProgram 2> error.log && \
$(call successCommand) || \
$(call failureCommand)
endf
This can't work as you have it here. Make doesn't contain an internal shell interpreter: make invokes a separate shell process. So, you can't jump back and forth between "shell code" and "make code".
Before make invokes the shell it will expand all make variables and functions. Then once the expansion is complete, the resulting string is passed to the shell and the shell runs the entire thing. Then once the shell is complete and exits, make looks at the exit code to see if the command succeeded or not.
So in your case, both $(call successCommand) and $(call failureCommand) will be expanded first, before the shell is invoked to determine whether ./myProgram succeeded or not.
Since you didn't give us any idea of what the successCommand and failureCommand do, there's not much help we can give beyond, if the test runs in the shell then the condition AND the results also need to run in the shell, and you can't use make functions like $(call ...).
I have a C program which uses argv[0] inside the program. I understand that argv[0] is the path of the program being executed. I want to pass a custom string as argv[0] to the program instead of its program name. Is there a way to do this in shell?
I read about exec command. But I am unsure about the usage. help exec says I have to pass exec -a <string>
Is there any other way of doing this?
Is there any escape method which I need to use if I am passing special characters or path of another file using exec command?
To clarify the problem:
I am running a program prog1. To enter a particular section in the program I have to give a SIGALRM to the program. This step itself was difficult as I had to create a race around condition to send the signal right when the program starts.
while true;do ./prog1 2; done & while true; do killall -14 prog1; done
The above while loops help me to enter the part of program and that part of program uses argv[0] for a system call. This system call is system(echo something argv[0])
Is there a way to modify the above while loop and put ;/bin/myprogram instead of argv[0].
Bottom line: I need /bin/myprogram to be executed with the privilege of prog1 and it's output.
exec -a is precisely the way to solve this problem.
There are no restrictions that I know of on the string passed as an argument to exec. Normal shell quoting should be sufficient to pass anything you want (as long as it doesn't contain embedded NUL bytes, of course).
The problem with exec is that it replaces the current shell with the named command. If you just want to run a command, you need to spawn a new shell to be replaced; that is as simple as surrounding the command with parentheses:
$ ( exec -a '; /bin/myprogram' bash -c 'echo "$0"'; )
; /bin/myprogram
The brute-force method would be to create your own symlink and run the command that way.
ln -s /path/to/mycommand /tmp/newname
/tmp/newname arg1
rm /tmp/newname
The main problem with this is finding a secure, race-condition-free way to create the symlink that guarantees you run the command you intend to, which is why bash adds a non-standard -a extension to exec so that you don't need such file-system-based workarounds.
Typically, though, commands restrict their behavioral changes to a small, fixed set of possible names. This means that any such links can be created when the program is first installed, and don't need to be created on the fly. In this scenario, there is no need for exec -a, since all possible "virtual" executables already exist.
I have a very basic problem using GNU Make 3.81 on Windows, I must be doing something very silly and I'm sure someone here will point it out in milliseconds. My problem is with using ";" to run multiple commands in the same shell.
As I understand it, make runs each line in its own command shell and so if you want to run two commands, one after the other, you must put them on the same line separated by a semicolon. In it's simplest form:
all:
echo hello; echo hello
...should produce the output:
hello
hello
But for me it produces the output:
hello; echo hello
In other words, the semicolon is being passed straight through to the shell, which doesn't make too much sense for cmd.exe.
I'm now ready to be embarrassed by everyone pointing out where I've gone wrong...
FYI, the reason I need this is that I'm using a $(foreach) loop which must execute two shell commands for each iteration.
You are be under the impression that ; is a GNU-make operator for executing multiple
commands in the same shell within a recipe. Not so. It is linux shell operator
for punctuating a sequence of commands on the same line. It is not an operator for
the Windows shell, cmd, so when the recipe:
echo hello; echo hello
is executed by make on Linux, it has the output you expect, but when executed by make
on Windows it just means echo this:
hello; echo hello
So, the answer is that your shell is the thing that has to understand that ; separates multiple commands on the same line, it's nothing to do with make. This is not the case for Windows cmd.exe but is presumably the case for the shells that normally arrive with environments that use make (Linux, msys etc.). In my case, a good workaround was this:
define useDef
echo hello
echo hello
endef
all:
$(call useDef)
With this form of "single-lined" definition I can invoke a multiline shell command inside $(foreach). Make still does each "hello" in its own shell but in my case that's OK because I'm appending outputs to a file. If you need two commands to be run in the same shell for some reason then, on Windows, you would need to write a separate batch file (which I suppose you could create from inside the makefile).
I know this question is relatively old, but I've stumbled across the same problem recently. The solution (for me) was quite simple. I replaced ; with &.
Basically
all:
echo hello & echo hello
will produce
hello
hello
in cmd.exe.
And it works with $(foreach) loops as well.
UPD: You also can use && instead of & if you don't want your commands to fail silently.
CMake's execute_process command seems to only let you, well, execute a process - not an arbitrary line you could feed a command shell. The thing is, I want to use pipes, file descriptor redirection, etc. - and that does not seem to be possible. The alternative would be very painful for me (I think)...
What should I do?
PS - CMake 2.8 and 3.x answer(s) are interesting.
You can execute any shell script, using your shell's support for taking in a script within a string argument.
Example:
execute_process(
COMMAND bash "-c" "echo -n hello | sed 's/hello/world/;'"
OUTPUT_VARIABLE FOO
)
will result in FOO containing world.
Of course, you would need to escape quotes and backslashes with care. Also remember that running bash would only work on platforms which have bash - i.e. it won't work on Windows.
execute_process command seems to only let you, well, execute a process - not an arbitrary line you could feed a command shell.
Yes, exactly this is written in documentation for that command:
All arguments are passed VERBATIM to the child process. No intermediate shell is used, so shell operators such as > are treated as normal arguments.
I want to use pipes
Different COMMAND within same execute_process invocation are actually piped:
Runs the given sequence of one or more commands with the standard output of each process piped to the standard input of the next.
file descriptor redirection, etc. - and that does not seem to be possible.
For complex things just prepare separate shell script and run it using execute_process. You can pass variables from CMake to this script using its parameters, or with prelimiary configure_file.
I needed to pipe two commands one after the other and actually learned that each COMMAND of the execute_process is piped already. So at least that much is resolved by simply adding commands one after the other:
execute_process(
COMMAND echo "Hello"
COMMAND sed -e 's/H/h/'
OUTPUT_VARIABLE GREETINGS
OUTPUT_STRIP_TRAILING_WHITESPACE)
Now the variable GREETINGS is set to hello.
If you indeed need a lot of file redirection (as you stated), you probably want to write an external script and then execute that script from CMakeLists.txt. It's really difficult to get all the escaping right in CMake.
If you can simplify your scripts to one command generating a file, then another handling that file, etc. then you can always use the INPUT_FILE and OUTPUT_FILE options. Or pass a filename to your command for the input.
It's often much cleaner to handle one file at a time. Although I understand that some commands may need multiple sources and destinations.
This won't work for me. I want to make some substitution and assign it to a variable in Makefile. An example is as follows but I prefer to do it with Perl since other substitutions can be more complex than this.
eval.%:
# make eval.exp-1.ans
# $* --> exp-1.ans
folder=`echo $* | sed -e 's/\..*//g'`
# OR
folder=`echo $* | perl -ne 'm/(.*)\.ans/; print $$1'
# I want that folder will be exp-1
echo $$folder ${folder}
Why this does not work? How can I do this kind of things in Makefile?
Your question is not clear. Are you trying to set a variable in your makefile, so that other recipes, etc. can see it? Or are you just trying to set a variable in one part of your recipe that can be used in other parts of the same recipe?
Make runs recipes in shells that it invokes. It's a fundamental feature of UNIX that no child process can modify the memory/environment/working directory/etc. of its parent process. So no variable that you assign in a recipe (subshell) of a makefile can ever set a make environment variable. There are ways to do this, but not from within a recipe. However that doesn't appear (from your example) to be what you want to do.
The next thing to note is that make runs each logical line of the recipe in a different shell. So shell variables set in one logical line will be lost when that shell exits, and the next logical line cannot see that value. The solution is to ensure that all the lines of your recipe that need access to the variable are on the same logical line, so they'll be sent to the same shell script, like this:
eval.%:
folder=`echo $* | perl -ne 'm/(.*)\.ans/; print $$1' || exit 1 ; \
echo $$folder
Have you tried the $(VARIABLE:suffix=replacement) syntax? In your case, $(*:.ans=). That will work for any suffix, even if it doesn't start with a dot.