add functions <signal.h> , getenv/system (node.js)<stdlib>

[ghost]

---

library.js , add system

---

i have updated in my branch library.js
to implement : system <stdlib.h>
it works on node.js
usage :

system ( "dir" ) ; system ( "node -v" ) ;
system ( "dir c:\\windows" );
system ( "em++ node001-getenv.cpp -o main.js" );

---

emscripten.h , add getenv

---

i have addedd getenv command <stlib.h>,,
it work on node.js
usage :

 char *s = getenv ( process.env.APPDATA );
printf ( "\n NODE.JS <%s> " ,s ) ; printf ( "\n NODE.JS <%s> " , getenv ( process.env.HOMEDRIVE ) ) ;
printf ( "\n NODE.JS <%s> " , getenv ( process.env ) ) ; 

---

emscrpiten.h, added pause

---

raise SIGSTOP and SIGCONT;
usage :
pause(1000);

---

emscripten.h, added MACRO

---

__SIGRTMIN
__SIGRTMAX

to implement other two function :

__libc_current_sigrtmax
__libc_current_sigrtmin

---

library.signal.js

---

addedd :

signal , raise , pause ,
sigprocmask , __libc_current_sigrtmax ,
__libc_current_sigrtmin

---

example

---

add example, for this command , in tests :

node001-getenv.cpp
node002-system.cpp
signal001-signal-raise.cpp
signal002-pause.cpp
signal003-sigprocmask.cpp

[kripken]

our convention has the { on the same line

[ghost]

ok i can repost you a code align ,with emscripten guide lines
{ in the same lines,

[ghost]

i need you ask a question : about free() . I have seen that this function
is not implemented yet, but counterpart in library.js _free, works good .
are there issues to implement correctly ?

[kripken]

free is implemented in system/lib/dlmalloc.c.

[ghost]

you're right
i have removed , getenv from emscipten.h

[ghost]

yes, but if you try this code, free does not work :

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <emscripten.h>
int main ( void )
{
// ************************************** free do not free
char *s1 = (char*) malloc ( 256 );
strcpy ( s1,"Hello\0" ) ;
puts (s1);
free(s1);
puts(s1);
// ************************************** free do not free
char *s2 = (char* )EM_ASM_INT ( {
var p = Module._malloc(256);
setValue ( p + 0 , 65 , 'i8' ) ; // A
setValue ( p + 1 , 66 , 'i8' ) ; // B
setValue ( p + 2 , 67 , 'i8' ) ; // C
setValue ( p + 3 , 0 , 'i8' ) ;
return p ;
} , NULL );
puts(s2);
free(s2); // do not free
puts(s2);
// ************************************** _free do free
/*
EM_ASM_INT ( {
return _free ( $0 );
}, s1 ) ;
EM_ASM_INT ( {
return _free ( $0 );
}, s1 ) ;
*/
puts(s1);
puts(s2);
char * s3 = (char*) EM_ASM_INT ( {
var str = 'ciao' ;
var ret = allocate(intArrayFromString(str), 'i8', ALLOC_NORMAL);
return ret ;
}, NULL ) ;
puts( s3 ) ;
free(s3); // do not free
puts( s3 ) ;
// ************************************** _free do free
/*
EM_ASM_INT ( {
return _free ( $0 );
}, s3 ) ;
*/
puts( s3 ) ;
return 0 ;
}

[kripken]

I don't understand, what do you mean by "free does not free"?

[ghost]

the command ' free ' does not release the allocated memory

2015-12-21 20:22 GMT+01:00 Alon Zakai notifications@github.com:

I don't understand, what do you mean by "free does not free"?

—
Reply to this email directly or view it on GitHub
#3948 (comment).

[ghost]

Happy Christmas Kripken !

2015-12-21 20:55 GMT+01:00 claudio daffra daffra.claudio@gmail.com:

the command ' free ' does not release the allocated memory

2015-12-21 20:22 GMT+01:00 Alon Zakai notifications@github.com:

I don't understand, what do you mean by "free does not free"?

—
Reply to this email directly or view it on GitHub
#3948 (comment).

[kripken]

Sorry for the delay here.

I still don't follow. How do you know it isn't freeing the memory?

[ghost]

excuse me, for my bad english
i'll post an issue soon
with more clarifications
regards Claudio

2016-01-07 1:54 GMT+01:00 Alon Zakai notifications@github.com:

Sorry for the delay here.

I still don't follow. How do you know it isn't freeing the memory?

—
Reply to this email directly or view it on GitHub
#3948 (comment).

[kripken]

This PR is quite old, closing. Feel free to reopen if still relevant.

[IngwiePhoenix]

Pretty sure I'll be poking a dead issue here.

I am poking around with WebAssembly and I want to access variables from the real environment - well, not me specificaly, but SWIG would like to. I run it via Node.

Studying the settings.js file containing several of the -s options, I came up with this configuration command:

$ /opt/emsdk/upstream/emscripten/emconfigure ./configure \
CC=/opt/emsdk/upstream/emscripten/emcc \
CXX=/opt/emsdk/upstream/emscripten/em++ \
--without-pcre \
LDFLAGS="-s ASSERTIONS=1 -lnodefs.js -s EXIT_RUNTIME=1 -s ALLOW_MEMORY_GROWTH=1 -s ALLOW_TABLE_GROWTH=1 -s ENVIRONMENT=node -s NODERAWFS=1 -s SUPPORT_ERRNO=1"

So far, so good. Passing environment variables though does not work unfortunately. Has a streamlined method of using environment variables been introduced in the meantime? Thanks!

[sbc100]

Emscripten seeds its environment from the ENV JavaScript variable. This works the same on the web as under node.

If you would like to node-specific behaviour to mirror the real environment, and that is genuinely useful, and I think we would accept such a patch. But these change of changes that are node-specific can also hurt portability so we should consider them carefully.

[IngwiePhoenix]

Why not go for an option like -s NODERAWFS=1 - but for the environment? This option already makes it very obvious that it is node-specific. Alternatively, would there be a way to "link" this behaviour in? I saw the pre-/post-js but don't have a full idea of how I could use those to essentially for(var k in process.env) { ENV[k] = process.env[k]; }.

[sbc100]

Sure either of those approaches would work.

That for loop in pre-js should work yes.

[IngwiePhoenix]

Had no luck so far. But I did find this:

functiongetEnvStrings(){if(!getEnvStrings.strings){// Default values.// Browser language detection #8751varlang=((typeofnavigator==='object'&&navigator.languages&&navigator.languages[0])||'C').replace('-','_')+'.UTF-8';varenv={'USER': 'web_user','LOGNAME': 'web_user','PATH': '/','PWD': '/','HOME': '/home/web_user','LANG': lang,'_': getExecutableName()};// Apply the user-provided values, if any.for(varxinENV){env[x]=ENV[x];}varstrings=[];for(varxinenv){strings.push(x+'='+env[x]);}getEnvStrings.strings=strings;}returngetEnvStrings.strings;}

The way pre-js is positioned, there is no viable solution to modify getEnvStrings at all to just reference process.env. In fact, if this behaviour was changed to something like this:

if(ENVIRONMENT_IS_NODE){getEnvStrings.strings=process.env;}else{/* ...implement the "standard" env as seen above. */}

then getenv and friends would happily read and respond with the actual system environment.

Looks like my best bet at this point is to look where this is generated and how I can change/update that and maybe make it a PR or something. But for now, I am a little lost as to how I can just add this feature without neccessarily modifying the Emscripten source.

[IngwiePhoenix]

Well, I did find a method. Hacky, but for now, works.

LDFLAGS appended: --pre-js (pwd)/env-init.js -s INCOMING_MODULE_JS_API="['preRun']"
env-init.js:

Module.preRun=functioncustomPreRun(){ENV=Object.create(process.env);}

This allows this program to display the proper PATH:

#include<stdio.h>#include<stdlib.h>intmain() { puts(getenv("PATH")); }

Not elegant but working. Wish there was an easier way to get this working, considering it's literally a one-liner. XD