Encrypting source code is not something that can be done in v8 due to
Function.toString and how v8 stores code for recompilation. The source will
be present in v8's heap somewhere. However, if all JS is encrypted with a
secret key and that must be provided before v8 starts it is possible to
See : https://github.com/isaacs/node-semver/blob/master/test.js
Does not have all the '+' checks you may want, but on glance it looks like
it would be fine except build.2 to build.11 (anything after + is lexically
sorted, semvers are not recursive by definition)
--
--
Job Board:
Arch specific registries using multiple registry support is much simpler.
npm --registries=ubuntu.x64.prebuilt.local registry.npmjs.org
when requesting against ubuntu.x64.prebuilt.local a proxy would prebuild
the module if it 404s. end. Can be done as an internal proxy as well.
--
--
Job
I am very wary of the module system needing to load modules, the less
external dependencies the better. That being said, you could use the built
in util functions, but then if a user application modifies the util
functions it will cause problems in module.js; tradeoffs.
--
--
Job Board:
Correct, however 2 pieces of bindings are used in node's case (dlopen and
fs for reading files/dirs). The rest is all in Javascript.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message
Also there is a shim for re-creating this in user
space: https://github.com/bmeck/node-module-system , see example for
`node`'s system.
On Thursday, April 4, 2013 2:29:20 PM UTC-5, mscdex wrote:
On Apr 4, 2:45 pm, Blaž Sovdat blaz.sov...@gmail.com wrote:
I'd like to know where in the
The one downside I have noted with node-gyp is how much of a pain it is to
bundle some package as a dependency that includes a different build manager
like cmake. Perhaps that is the situation here.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
My main problem is not with JS but having native modules break if we do
this. V8's C++ API is significantly easier to use than the other JS VMs
when I have done experiments, however, that C++ API makes some assumptions
that are not always true across JS Core, Nitro, or the Monkey engines.
I think this is a little out of scope for node-core, some test runners for
example do automatically include various bootstraps and/or globals (chai in
particular comes to mind). Having this in included as a CLI flag would need
to have some serious points about why this would be beneficial over
When we have been making large scale software a few things have come up.
1. Callbacks suck for linear workflows (a-b-c)
This kind of workflow results in what I like to call the mudslide.
Your code really wants to expand into some minor nesting once you have
shared state (sometimes immutable
Domains are useful to categorize and debug the error, even if you shutdown,
don't just throw them away.
For performance starved JS (talking working under 50MB in a production
area) idempotentcy / stateless arch generally results in memory usage or
offloading to a single authority. Single
Manually combining state is something I do not want to give up. Passing by
argument is my preferred method rather than introducing variables
(async.waterfall). All compiler based syntaxes I have seen give me concerns
about shared cache if multiple things are done in parallel and you use
shared
https://gist.github.com/bmeck/5287456
Feel free to spec out a reference problem set by forking and modifying.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed
This should also specify if you are using Windows or not when making
the decision. Native modules on Windows are less common due to the
semi-long setup needed to make/secure a build server for me. On *nix
modules are amazingly useful, but on windows I usually end up offloading
complex native
If you just want to call into C https://github.com/rbranson/node-ffi
or
If you actually need to deal with the event loop (such as doing things in
multiple threads and queueing them on the loop):
http://nodejs.org/api/addons.html
On Friday, March 22, 2013 11:10:49 AM UTC-5, Gaurav Meena wrote:
Did npm-debug.log say anything? Can you gist that output?
On Thursday, March 21, 2013 4:13:57 AM UTC-5, Christian S. wrote:
Hi,
I always get the return code 7 when I run npm search term from python
( subprocess.check_output([ 'd:/nodejs/npm.cmd', 'search', 'connect' ],
shell=True) ). When
Can't seem to click on the link (404ish style page saying missing or not
shared).
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups nodejs
https://github.com/joyent/node/issues/1167 - stdio options need to be
specced out still.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups
Most real world use cases for this are not for performance, they are for
shelling out to utilities, the alternatives are hideous as we found in nsh
: https://github.com/AvianFlu/nsh/wiki/nsh
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
For reference to those who are taking this seriously: multiple comments
asking for sources/examples on many issues brought up (in video and
comments) should lead to skepticism about many of his claims in the
comments.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
Unless you really need to script out SNI and/or NPN negotiation in your
balancer, termination via `stud` or something else is recommended on 0.8.x
unsure on 0.10.x.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
Congratulations on going public :). Always good to see the Node community
thriving.
Cheers,
Bradley
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the
You should check out module-smith; let me know if you need help setting it
up properly. Make tarballs / zips and distribute them directly as
precompiled binaries.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
Well right now the cache (~/.npm) does follow this but does local copies
instead of linking when using `npm install`. Linking is only going to
happen with `npm link` so that nothing in a project can change out of band
unless it was explicitly linked.
--
--
Job Board: http://jobs.nodejs.org/
This structure is currently present in node.
* NPM uses local modules in a similar structure with `/usr/lib/node`
replaced with the directory containing your `package.json`.
* Global modules are installed as local modules name-spaced to the
appropriate directory such as `/usr/lib/node_modules`.
There are some projects on this already such as `nexe`. If you want to talk
about it in depth feel free to ping me on irc.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are
I think both `nexe` (dont quote me im not on project) and the private
versions I know of are looking to windows as the main usage for bundling.
NJ is looking into security around bundled executables before we do
anything public.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
Very interesting. Looks like some good things for us people who are used to
vanilla node projects far more than meteor.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are
We are doing some dark magic upcoming
using https://github.com/bmeck/node-module-system +1 if we can get a
standardized require shim. Does not need to be ours.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You
Generally I abstract out my implementations and my interfaces. This means a
more verbose code base, but allows me to have the renderer inserted via
dependency injection. It is independent of bundling tech and I have used it
with browserify as well as with AMD to great success. Decoupling those
Loading in modules can be time consuming for one off apps or resource
constrained apps. This only requires them when they are being requested.
For long running apps that can consume time and resources at startup this
is generally not needed / wanted.
--
--
Job Board: http://jobs.nodejs.org/
Right now without compaction you are looking at 45~ GB of data it looks
like on the current npm public registry, and it can be replicated without
needing any permission.
On Monday, February 18, 2013 1:01:33 PM UTC-6, andy wrote:
Cool, thanks for the suggestion(s).
Two quick questions:
1)
Any reason not to use http-proxy and passport without express?
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups nodejs group.
To post to
A long time ago I built a proof of concept for something like
this: https://github.com/bmeck/kitsune
After thinking about maintainability I decided to avoid doing this sort of
behavior but left the code up as an example.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
GET the document from CouchDB, change the row appropriately, PUT it back to
CouchDB. All using HTTP. CouchDB will complain about merge problems if the
_rev field does not match, if so, retry.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
Proxies are a turtles all they way down approach for things like this,
don't return the actual property, return a Proxy to the property that can
have reference to the path used to get there. However, this ruins
performance, and cannot represent primitives as values. A function for full
path
I really like proxies, but they are not a cure all; in some situations they
are very very beneficial (NodeList in the DOM spec for example), but they
do not like valued tree structures.
But yes a function that delegates out to a tree structures is what I would
recommend.
--
--
Job Board:
use xargs and -e instead of a direct pipe.
On Wednesday, January 30, 2013 2:36:04 PM UTC-6, Kilian C. wrote:
using /dev/stdin as file reference kinda works
$ echo console.log(process.argv) | node /dev/stdin --myarg
[ 'node', '/dev/stdin', '--myarg' ]
but it messes with the node_modules
Are there any conventions when sharing data between 2 mumps databases?
Format or otherwise?
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Any errors in the interpreter/compiler could have impact on peoples well
being, I would advise against doing this on old code that has been
functional for long periods of time and may rely on edge cases.
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
Dependencies on a resource is pretty fancy pantsy. I like it.
Reflected web admin trigger admin...
http://memegenerator.net/instance/33969625
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this
Good to see you around. Keep us up to date with any slides / videos :)
--
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups nodejs group.
To
Are we talking about the frame limitation here? Packets above a certain
size will be split into frames as appropriate.
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are
Austin
: https://github.com/joyent/node/issues/2486
https://github.com/joyent/node/issues/3042
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Alexey, depends on the system. Long story put short:
If you have persistent workers (say one per user not per request), this can
be beneficial in some situations where running user code on your machines
is less costly than sending things over network or for security you cannot
(ie. something
A better use of time is supporting SES ( https://github.com/drses/ses ). It
needs a legitimate CommonJS wrapper without globals and documentation on
usage. See bin/ses for an example.
As a side note: you will want --harmony support. Once you have an OS
protection layer and that you are in ok
Austin: won't go into great detail, but heres a fun little example:
```javascript
function exploit() {
console.log.constructor(process.exit(42))();
}
var result = require('vm').runInNewContext([
'use strict;',
'('+exploit.toString()+')()'
].join('\n'), {console:console});
console.log('I
See my earlier post with OS security and the various things about it. Even
in a child process you can wreak havoc. Also, you need to pass in some sort
of context or get some sort of output for you to do anything of value. It
needs to be made into a string, fire some function, return a value, or
Yes, for basic scripts gf3's module works fine, but still stick to OS
security as well in case a new JS exploit arises. The reasons it is ok is
as follows:
1. Communication uses serialization in a safe runner generated inside the
context that is in strict mode. (General idea from #7)
2. RPC is
A while back I discussed the problems with the node-sandbox module, I wrote
a solution, but it broke on newer v8s, have not spent the time to track
down the problem. Even then serious care must always be taken to not allow
remote code execution. I know ways that horrify people using just
type
Hey Mark,
Is there a sandbox to test out SES in, or binary I can run with examples?
Looked in the SVN but did not see a Readme to get started with.
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this
What is `time node --prof x.js`, always check `time`. Also C++ level stuff
for in depth profiling use a C++ profiler, `gprof` for gcc etc.
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message
You can run it perfectly safely, but always understand you want the OS for
protection. That is the first line, not a module.
Now running something with shared code... harder, but SES shows promise and
does what I mentioned using wrappers; hopefully, we will see it usable in
both browser and
writing
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups nodejs group.
To post to this group, send email to nodejs@googlegroups.com
To
Profile the machine, 51 elements means very little overall. The elements
could vary in size from a vast number of reasons (the simplest being you
have strings of different sizes). You could also have a daemon watch RSS
usage of the machine to react accordingly while you profile by making
If you need to break something off as to not interfere for w/e reason, look
at `process.nextTick`, if you need to chain together a reaction to an event
from async functions look at `async` from https://github.com/caolan/async
or `understudy` from https://github.com/bmeck/understudy depending on
Completely agree that not having the default action at the end would cause
issues in a single middleware stack.
I use middleware trees rather than single stack with Understudy, it would
go to the end of an action (for example the default action such as just
closing the HTTP request). I run
I am not sure I understand the use of a pull stream in this context. Can
you explain in a small example of how to use a stream instead of
domains/middleware/etc. in this fashion for control flow?
On Tuesday, January 8, 2013 3:59:29 AM UTC-6, Raynos wrote:
Sounds like you want a pull stream to
Anyone have a sane and performant way that they break out of a domain or
action such as middleware? I am trying to design such behavior but am faced
with a multitude of interesting issues. The classic example of this problem
can be seen with Array.prototype.forEach (though this has the easier
This is good, gets me thinking. This brings up that there are 2 styles of
approaching this issue:
abort style - manually cause the action to fast fail
condition style - provide a condition to check on every step of the action
May help others thinking about this as well to think of both
What are the security features to prevent malware installation using this?
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups nodejs group.
To
Well if its just runtime needs, you could throw an error on require if a
global for w/e you need is not present? Placing Javascript dialect into the
package.json would get complex and hairy quite quickly.
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
If you are using a non-standard (harmony) expectation in your library you
should test and bail on an environment that does not supply the interfaces
expected. I think throwing on missing expectations is one solution the
other is to require a shim if the expectation is not present. Both are
Fixed using nextTick in
https://github.com/bmeck/node-tart/blob/b44c960b18904b4ba1b36f02c2e3f569af948503/lib/substream.js
.
Basically, when having a stream that has read the end of it's data and is
also ending at the same time, you need to nextTick the cb(null, null)
otherwise the consumer
Debug:
At least give a way to disable it for when im doing live debugging and want
to regain the port and debug a different process.
On Thursday, December 27, 2012 10:51:44 PM UTC-6, Isaac Schlueter wrote:
On Thu, Dec 27, 2012 at 6:52 AM, Joel Brandt jobr...@adobe.comjavascript:
wrote:
I have a situation where I want a sub-stream (in same manner as a
sub-string) of a ReadableStream to be treated as a different stream
entirely.
Currently I have the following:
```
//
// Create a stream that consumes a future part an existing stream
//
var ReadableStream =
Personally I did not want to write a buffering system for in memory _read
and just used a PassThrough stream for node-tart's substream parts
: https://github.com/bmeck/node-tart/blob/master/lib/substream.js . I'm
trying as well to find a sane way to allow internal buffering from a
different
Very interesting. Seems to still have some graphical jitters, but could be
very promising.
Cheers,
Bradley
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the
Encapsulated bindings.
Make all bindings available w/ some naming convention for the namespaces.
Have a constructor function per type
* Have Animal.prototype be the members of Animal::
* Have Dog.prototype be the members of Dog::
Attach the static functions to constructor functions
Add sugar if
I think the main reason people encourage others to compile to JS is that it
can be a prepublish script to npm (you don't have to change .coffee to .js
manually every publish) and in doing so you do not require 'coffee-script'
to be loaded by the module that uses yours, making requirements of
Depends, if a module holds a reference to the module being reloaded and
there is a shared cache being used by that module. You would still end up
with 2 references to different things. I would need to know way more about
whats trying to be done / API to make any sane comment on if it looks like
If the module uses some form of internal storage (for caching, module level
object registration, etc.) it would be duplicated and then you could see
some interesting cache mismatch bugs.
For example in the following module 2 registries would exist rather than
the expected one if completely
Choose either, I do things in callback style personally to match the
ecosystem when I write libraries, thats the expectation of the ecosystem.
For application logic, do whatever makes it faster. There are exactly 0
absolute wins on either side. I can go on at length about the reason both
have
Correct but people use factory functions based on string lookups in most of
these modules for some forsaken reason (udp/tcp do not have same api
basically).
On Friday, November 9, 2012 1:32:39 PM UTC-6, Matt Sergeant wrote:
Yeah that was the one. I thought mDNS was just DNS packets sent over
Yep, but there are interesting ways to do things still, like STUN and SOCKS
bindings if you have a proxy of your own out there you can generally setup
the tunnel inside most PaaS. Generally though the PaaS is stuck to
preallocating port/host pairs and charging for them since most don't give
Having worked on load balancers I can say load balancing multiple apps on
raw TCP is less than trivial. Basically you end up having to make users
declare their socket needs before hand, and you assign a host:port that
will map to their app inside the internals of the PaaS. If someone has good
This sounds like a shared state lookup. Document and specify an interface
for the renderers (webgl and 2d in this case). Once you have done that,
in getContext
Surface.prototype.getContext = function (name, options) {
var renderer = this._renderers[name] // _ prefix is for implementation
Posix has the ability to have hard links on disk where they have the same
inode but different paths, unsure if this affects Window's junctions in the
same way though.
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
Take it to a different topic if it becomes unrelated also.
On Wednesday, October 24, 2012 1:15:49 PM UTC-5, Marco Rogers wrote:
Can you elaborate on domains just feel wrong. I'm interested in how the
domains api comes across. I've expressed my concerns with it in the past.
:Marco
On
1. npm dedup can help.
2. if that is not enough: all modules can reference parent module's
node_modules, so if you recursively move all dirs into the root
node_modules it should work.
a/node_modules/b/node_modules/c - a/node_modules/c
On Wednesday, October 17, 2012 1:07:54 PM UTC-5, Justin
We at Nodejitsu are releasing a part of our build server, Module Smith.
It is the NPM wrapper we are using after we have stripped out all of our
fancy logic. As with most build servers, the permissions you run this under
should be taken into consideration.
We have stripped out a lot of our
Link may be important:
https://github.com/nodejitsu/module-smith
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups nodejs group.
To post to this
that would do
that.
On Thursday, October 11, 2012 9:30:12 PM UTC+2, Bradley Meck wrote:
you can do something like
```
npm i cat package.json | json -e this.bundledDependencies='$(for FILE
in $(ls -A node_modules); do printf $FILE ; done)'.split(/ /)
package.json npm pack
```
We do
Isolates were taken a look at a while ago. They are similar to what you
want, but still unable to do so. It would generally be more sane to just
scale out processes was what was decided. The edge cases of giving multiple
light weight isolates proved to be very difficult to prevent Isolates
My problem is not with using Travis, it is a warning. Travis does sometimes
fail, and we have had situations where commits cause tests to fail
according to travis while all local tests run on all machines we test. The
cause of this has generally been the problems mentioned.
On Friday, October
Probably true especially regarding code complexity, but disk space is
cheap. Though, the time to install all that stuff is a real bummer. If `npm
install` was faster, would this be less of an issue?
Disk is cheap, but if you actually do end up loading all the libraries at
once you can
https://npmjs.org/doc/config.html#rebuild-bundle
On Thursday, October 11, 2012 8:49:10 AM UTC-5, Mariusz Nowak wrote:
Daniel, I think 'npm rebuild' may work for installation of
already created bundle, but still what's missing is how to download all
dependencies into correct folders without
Forgot to mention, bundle the dependencies on your build server.
On Thursday, October 11, 2012 10:25:56 AM UTC-5, Bradley Meck wrote:
https://npmjs.org/doc/config.html#rebuild-bundle
On Thursday, October 11, 2012 8:49:10 AM UTC-5, Mariusz Nowak wrote:
Daniel, I think 'npm rebuild' may work
bundle that was already
prepared, but do you know the way how can I create such bundle? This is
what I'm missing (!) :)
On Thursday, October 11, 2012 5:26:24 PM UTC+2, Bradley Meck wrote:
Forgot to mention, bundle the dependencies on your build server.
On Thursday, October 11, 2012 10:25
Just beware of Travis CI failing for situations that are beyond your
control. Missing C libraries, OS issues, and external resource needs can
all be problematic. Also I have noted at least in the past sometimes travis
fails to provision VMs appropriately. Running tests in your environment is
a. who is actually messing with crypto after the fact. I would like to know
the reasons to do so.
On Monday, October 8, 2012 6:24:36 PM UTC-5, Isaac Schlueter wrote:
Currently, the crypto module defaults to using 'binary' encoded
strings everywhere as the default input and output encoding.
As long as NPM works, since it uses Node's module loader apparently, I am
sure some horrible but interesting bridges will be crossed on the NPM
registry. Looking forward to it.
On Thursday, October 4, 2012 12:16:24 PM UTC-5, Jonathan Buchanan wrote:
I'm at JavaOne, for my sins, and I've been
split is a fairly expensive operation, for the most part I would guess the
regex compiler would do a better job and avoid GC fluff. Ugly though.
http://jsperf.com/ipv4-regex
--
Job Board: http://jobs.nodejs.org/
Posting guidelines:
Look into domains. You can just add a property to process.domain was
something that was discussed at NodeConf summer camp.
On Thursday, September 20, 2012 9:07:23 AM UTC-5, David Boon wrote:
Forgive me for reviving this debate about threadlocals. Before anyone
loses their heads, I'm not
There are long discussions in the Node community about what happened when
we did try to use Isolates. The lack of thread level protections from
things like mucking with process.* and the fact that native modules need to
add complex support for Isolates to be first class (many C level libraries
the approach taken by threads a gogo. I am not
against threading, just very cautious.
On Sunday, September 16, 2012 6:11:48 PM UTC-5, rektide wrote:
Thanks so much for the informative reply: that's a lot of things I don't
know about!
On Sunday, September 16, 2012 6:22:50 PM UTC-4, Bradley Meck
How do people unload entire modules after they are done using them, we have
some very long running apps we are trying to reduce the footprint on and
unloading certain modules seems to be a good start. For example if we could
eventually end up with something like:
require.unload('nconf') after
3
Hook into .on('request') or .on('connection') if you want it, thats how i
do it already.
On Wednesday, August 29, 2012 6:07:23 PM UTC-5, Isaac Schlueter wrote:
Do you use the clientError event?
Please choose one:
1. Yes, I use it all the time. Please don't change it ever, thank you.
try setting the `npm_config_arch` environmental variable to your desired
architecture when spawning `npm`.
Related Code:
https://github.com/TooTallNate/node-gyp/blob/master/lib/node-gyp.js#L142
https://github.com/TooTallNate/node-gyp/blob/master/lib/configure.js#L235
On Tuesday, August 28, 2012
I have heard of a different person with what sounds like a similar problem,
but no idea how wide spread it is.
On Friday, August 17, 2012 2:00:49 PM UTC-5, adrians wrote:
With node 0.8.7 x64 (Windows 7), npm seems to not see global modules
installed under userdir/npm/... It only seems to
1 - 100 of 133 matches
Mail list logo
