A better "shebang" would be

///bin/true; /usr/bin/c "$0" "$*"; exit $?

as it would keep file compileable with normal gcc invocation. See example of this here (also discussion on StackOverflow).

I got your script working on freebsd with multiple parameters using this as my first line

#!/usr/bin/env -S c "-I/usr/local/include/ -g -L/usr/local/lib/mysql/ -lmysqlclient --"

might be worth mentioning in the readme

Having the license in the same file as the readme makes this more difficult to package than it needs to be., as I would need to manually parse the readme file. If the license was in a separate file, I could just copy it into place.

!/usr/bin/env c

At the top of the script searches your path for 'c' and execute that script.

Means you can use it no matter where your 'c' is installed (~/bin/ /usr/local/bin or /usr/bin et al).

Hi, I noticed that depending upon what I pass to C sometimes result is not what I expect. Having looked at https://github.com/ryanmjacobs/c/blob/master/c#L54 I am a bit confused and thought it is worth raising this question. I see that here instead of trying to process all command line arguments, we there is a for arg in $1, which even if we do a shift later won't mean that this for look is going to run more than once. I assumed that maybe that should have been $@, but it would break some of this logic also.

Anyway, I think my question, since there are not comments there to help me understand, is whether or not this loop at line 54 was meant to be gone through more than once. If so, I think there may be problems with it.

Thanks!

root@niuyuling:/mnt/c/Users/niuyuling/Desktop/libini# ../c/c test.c -L./ -lini -static
/usr/bin/ld: /tmp/cc3AcSun.o: in function main': test.c:(.text.startup+0x21): undefined reference to getinikeystring'
/usr/bin/ld: test.c:(.text.startup+0x5f): undefined reference to getinikeystring' /usr/bin/ld: test.c:(.text.startup+0x9a): undefined reference to getinikeyint'
/usr/bin/ld: test.c:(.text.startup+0xdd): undefined reference to getinikeylong' /usr/bin/ld: test.c:(.text.startup+0x121): undefined reference to getinikeyfloat'
/usr/bin/ld: test.c:(.text.startup+0x170): undefined reference to putinikeystring' /usr/bin/ld: test.c:(.text.startup+0x18d): undefined reference to getinikeystring'
/usr/bin/ld: test.c:(.text.startup+0x1cd): undefined reference to `getinikeystring'
collect2: error: ld returned 1 exit status
root@niuyuling:/mnt/c/Users/niuyuling/Desktop/libini#

root@niuyuling:/mnt/c/Users/niuyuling/Desktop/libini# ../c/c 'test.c -L./ -lini -static'
/usr/bin/ld: 找不到 -lini
collect2: error: ld returned 1 exit status
root@niuyuling:/mnt/c/Users/niuyuling/Desktop/libini#

可不可以支持多文件或者链接库文件
Can I support multiple files or link library files?

just like this cli

sudo install -m 755 c /usr/local/bin/c

What is the license for this code? I'm interested in it, as I am the maintainer of awesome-c, and would prefer to have license terms to include.

Hi,

pretty useful tool. In order to make it work on MacOS I had to make this change:

-    size="$(du --block-size=1024 -s "$tmproot" | tail -n1 | cut -f1)"
+    size="$(du -k -s "$tmproot" | tail -n1 | cut -f1)"

The reason is that du on MacOS does not support --block-size. I can try to fix the script to detect MacOS and use du -k, but I noticed that this line has been edited in commit 91b66a0, so I don't know which is the best solution.

Thanks!

Giovanni

I'm just very curios to know how its possible that just by putting shebang, no 'c' prefix command requred. just run it directly like this:

$ chmod +x hello.c
$ ./hello.c
Hello World

its the feature of bash?

i cant find where you implemented this feauture.

What do you think about using

#!/usr/bin/env c

as shebang to support prefixed install. Then you can e.g. install in /usr/local and still run the examples:

PREFIX=/usr/local bpkg install ryanmjacobs/c

It would be awesome if we could pass comma separated RPATH's in the shebang.

The sed command that removes/replaces the shebang appears to fail on OSX yosemite:

$ c << EOF
#include <stdio.h>

int main()
{
  printf("Hello World!\n");
  return 0;
}
EOF
sed: 1: "/tmp/c.ox3.stdin.c": command c expects \ followed by text
Hello World!
sed: 1: "/tmp/c.ox3.stdin.c": command c expects \ followed by text

I found a "c" file in the "~/bin" directory of an old computer.

I figured what it was for pretty soon, took me a long while to find the source of the file :)

BSD install doesn't have the -D option, which causes installation to fail on systems that use this version of install.

My (probably incomplete) understanding is that this can be resolved by splitting this into a mkdir -p followed by install sans the -D option, but I don't know if there is a better way that works with the bpkg 'install' field.

For reference, the relevant install command: https://github.com/ryanmjacobs/c/blob/master/package.json#L6

The script does not work on files with paths containing whitespaces.

[jimmy@toolbox my folder]$ pwd
/home/jimmy/my folder
[jimmy@toolbox my folder]$ ./foo.c 
cc: error: folder: No such file or directory
[jimmy@toolbox my folder]$ c foo.c
cc: error: folder: No such file or directory
[jimmy@toolbox my folder]$ c "/home/jimmy/my folder/foo.c"
cc: error: /home/jimmy/my: No such file or directory
cc: error: folder/foo.c: No such file or directory
cc: error: folder: No such file or directory
cc: fatal error: no input files
compilation terminated.

hello man can you help me in C ?
i now how to extract ANCII chars and concaternate them..
what i need to know its how can i execute that ancii string after concaternation ..

example:
1º i what to extract the string whoami from ancii chart
2º concaternate the chars to build whoami command
3º and execute the string concaternated ..

this code does allmost everything..
it extracts the chars, concaternate them ...
but it only prints the string ..
but i dont what to print ..i want to execute ..

#include <stdio.h>
#include <string.h>

  int main(void)
    {
      /* variable declarations */
      char a = 119;  /* ancii character w */
      char b = 104;  /* ancii character h */
      char c = 111;  /* ancii character o */
      char d = 97;   /* ancii character a */
      char e = 109;  /* ancii character m */
      char f = 105;  /* ancii character i */
 
      /* concaternate and transform decimal values to ancii chars using putchar */
      //char command = putchar('a');putchar('b');putchar('c');putchar('d');putchar('e');putchar('f');

      /* execute command using system() */
      int system(char command);
        system(command);
    }

While caching is great in general, I found it to be hard if you are developing code in a header file. Is there a way to shut off the caching?

Hi, can we please get this on brew?

it would be great to have a --debug flag to launch gdb

Thanks for this tool. I've been looking for something like this for a while.

I'm running into an issue where, if I run c ./main.c, and the contents is something like this:

#include <stdio.h>

int main(void)
{
    printf("%s\n", "Hello, World");
    return 0;
}

I get the following output:

Hello, World
du: illegal option -- -
usage: du [-H | -L | -P] [-a | -s | -d depth] [-c] [-h | -k | -m | -g] [-x] [-I mask] [file ...]

Seems like --block-size in size="$(du --block-size=1024 -s "$tmproot" | tail -n1 | cut -f1)" isn't valid on Catalina.

Any ideas? #38 mentions this error.

In my Emacs config, I have clang-format do most of my formatting.

Adding a shebang: #!/usr/bin/c -Wall -Wextra -Werror seems to mess with the formatter.

Perhaps it might be possible to put the shebang in a comment:

// #!/usr/bin/c -Wall -Wextra -Werror

and have c look for it with regex: ^\/\/[\s]?#!.*?

Instead of always hashing to check if cache is invalidated, why not use stat or date to check if a script has been modified since last compilation? If the script has been modified, we can fallback to hashing, as usual

Hi @ryanmjacobs,

I just want to mention similar work. You might consider adding a link to your readme or taking inspiration from the ~/.c99shrc config file.

• https://github.com/RhysU/c99sh
• Other nice C stuff: https://notabug.org/koz.ross/awesome-c# (You might want to add your project to the list)

Currently we can't pipe stdin input into a C program, e.g.

#include <stdio.h>

int main(void) {
    char buf[1024];
    while (fgets(buf, 1024, stdin) != NULL) {
        puts(buf);
    }
}

$ c file.c arg1 arg2 < /dev/urandom
$ ./file.c arg1 arg2 < /dev/urandom

Because we use stdin for file input, I don't think we can also use it for program input as well. I'm not quite sure how to approach this... maybe we can only choose one or the other?

(btw, stdin input without piping works just fine)

$ c file.c arg1 arg2
hello
hello

world
world

^C

When I use Shebang as described in README.md, I get an error.

This seems to be a compatibility issue with FishShell at the moment, as it works fine with Bash and Zsh.

I checked the source code of FishShell and I think the problem is here https://github.com/fish-shell/fish-shell/blob/master/src/exec.cpp#L177

Currently still looking for a solution

We need a command that doesn't require gnu-sed to be installed with brew.

This should work on both GNU and BSD systems:

$ cat > file.c
#!/usr/bin/c blah blah blah
^D
$ sed -i.bak '1!b;s/^#!/\/\/#!/' file.c
$ cat file.c
//#!/usr/bin/c blah blah blah

However, I don't have access to a BSD system...
It works just fine on GNU. Can someone confirm that it works on BSD systems, too? Thanks.

It would be great to be able to pipe into c :

cat file.c | c

c < file.c

This would allow the use of tools that compile into c such as awib.