olegp / common-node Goto Github PK

Remove connect, as it adds additional overhead and may do caching.

https://gist.github.com/1754568

This caused problems with system.stdin before & is now causing issues with reusing the same stream for HTTP Keep Alive connections.

Hi,

Just wondering if there is a way to get something like Nodemon or node-supervisor to work with common-node, so that it is not necessary to restart the process to see JS file changes. I'm on Windows.

Thanks!

Ringo modules frequently use destructuring assignment, adding optional support for Traceur would allow for those modules to be used without modification.

... and bring it in line with the one in Ringo.

Hi,

I went through these steps:

1. Install Node for Windows from http://nodejs.org/ (MSI v0.6.17)
2. Restart computer
3. Run on command line: npm -g install common-node and wait for files to download and install (no noticeable errors).
4. Try various mutations of "common-node $NODE_PATH/common-node/examples/hello.js"

No matter what I do (removing $NODE_PATH, changing directory to the examples folder and just executing hello.js, etc.) it seems I get "The system cannot find the path specified."

This is in line with Node.js

events.js:72
        throw er; // Unhandled 'error' event
              ^
Error: spawn ENOENT
    at errnoException (child_process.js:945:11)
    at Process.ChildProcess._handle.onexit (child_process.js:736:34)

common-node/subprocess does this to the command line:

var arguments = command.split(' ');
command = arguments.shift();
var child = childProcess.spawn(command, arguments,...);

which isn't going to fly...

via @alexlamsl

Looks like on('socket') no longer provides a socket as a parameter.

TypeError: Cannot call method 'setTimeout' of undefined
    at HttpClient.read (httpclient.js:192:13)
    at HttpClient.finish (httpclient.js:222:24)

path.js:313
        throw new TypeError('Arguments to path.resolve must be strings');
              ^
TypeError: Arguments to path.resolve must be strings
    at Object.exports.resolve (path.js:313:15)
    at module.exports (/home/oleg/Projects/common-node/lib/run.js:39:23)
    at Object. (/home/oleg/Projects/common-node/bin/common-node:3:22)
    at Module._compile (module.js:456:26)
    at Object.Module._extensions..js (module.js:474:10)
    at Module.load (module.js:356:32)
    at Function.Module._load (module.js:312:12)
    at Function.Module.runMain (module.js:497:10)
    at startup (node.js:119:16)
    at node.js:903:3

/cc @alexlamsl

Ideally, I'd like to keep this as a single project/package so that RingoJS code such as the following would work as is:

require('system').exit(0);

Unfortunately Node's approach to resolving modules and proposed removal of exports.paths.push() makes this difficult. The options are:

• use exports.paths.push() to add the ./lib dir to Node path & worry about this functionality being removed later - question is where could this be called?
• create a single module which combines the exports from all the modules in this project - this will require changes to RingoJS apps (that being said, they will need to be changed anyway to work around lack of destructuring assignment in V8)
• break the project up into separate packages, one per module - this would allow for a subset of the modules to be used and the dependencies between modules to be more clear; however there will be clashes with other packages already in npm (e.g. binary) - also this wouldn't resolve the issue of fs being a reserved module name

If you have any other suggestions, please leave them in the comments below.

http://ringojs.org/api/master/ringo/worker/

... using Node cluster perhaps.

/cc @alexlamsl

Thank you to @shangaslammi for bringing this up. To see what's broken, run the unit tests.

We probably need more unit tests to get full coverage. References:

• http://wiki.commonjs.org/wiki/Filesystem - spec we are working against
• https://github.com/kriskowal/narwhal-lib/blob/master/lib/narwhal/fs.js - Narwhal's implementation
• https://github.com/ringo/ringojs/blob/master/modules/fs.js - Ringo's implementation

... by using https://github.com/visionmedia/node-jscoverage

http://ringojs.org/api/master/ringo/httpclient/

e.g. https://github.com/olegp/common-node/blob/master/lib/system.js#L82

/cc @alexlamsl

https://github.com/ringo/ringojs/blob/master/src/org/ringojs/wrappers/Binary.java#L291

(node) warning: possible EventEmitter memory leak detected. 11 listeners added. Use emitter.setMaxListeners() to increase limit

at EventEmitter.<anonymous> (events.js:123:17)
at new <anonymous> (/common-node/lib/io.js:26:14)
at receive (/common-node/lib/jsgi.js:78:15)

To reproduce, try to connect to a different port. The problem is likely in common-node's httpclient and due to the for(;;) loop, but could be libuv/node related as well.

Hi,

With JSGI, I don't see a way to set the port. Is there a way to change it from 8080?

Also, how do I get command line options in Windows?

Thanks!

As @kof pointed out, this is already in Node core.

Need to make sure the removal doesn't break any of the tests.

When node -v v0.5.4 & npm -v 1.0.17 are used the following issues were spotted by @dsowsy:

• npm root -g returned an error about using the deprecated require.paths
• NODE_PATH had to be manually added to ~/.bashrc
• fibers and traceur are installed in common-node/node_modules, however running tests gives error saying fibers cannot be found; installing them manually globally fixed this

Please try to reproduce this and comment here on how we could possibly address these issues.

Currently the whole process exits if an exception is thrown inside the request handler, should probably just have it return a 503 instead.

laverdet/node-fibers#72

works fine under v0.8.18 (Windows 32-bit)

node_modules\common-node\lib\io.js:88
                fiber.run();
                      ^
Error: socket hang up
    at createHangUpError (http.js:1360:15)
    at ServerResponse.OutgoingMessage._writeRaw (http.js:507:26)
    at ServerResponse.OutgoingMessage._send (http.js:476:15)
    at ServerResponse.OutgoingMessage.end (http.js:887:18)
    at send (node_modules\common-node\lib\httpserver.js:117:12)
    at exports.jsgi (node_modules\common-node\lib\httpserver.js:128:7)

seems like this could be a real winning framework combination but needs some better simple CRUD examples to get folks like us using it. I looked at this last night but really couldn't easily get going with a simple app.

we like the use of fibers - regardless of what folks on the node forum say, callbacks are not easy, and most developers prefer simple sequential programming. Hence, the popularity of Rails. Also really like use use of Mongo shell commands as in Mongolian. Also the use of simple templates like ejs and Mustache.

BUT, there needs to be some simple examples to get folks going with common CRUD type apps - I suggest the project needs the 5 min blog and to-do examples. Even a simple hello world with a form to input name and then a template that reports "Hello " with a typical directory structure.

https://groups.google.com/d/msg/nodejs/d_8EL8fMRW4/jP6H-5IuSKwJ

http://olegp.github.com/common-node/graphs/buffer-alloc.png

Results are far worse than other benchmarks, most likely due to the slow zero() function. Should try using clear from https://github.com/bnoordhuis/node-buffertools/ or drop the requirement to clear the buffer (this would break tests though).