Fixes issue commented on in #377 #501

einaros · 2011-08-31T13:21:14Z

See the following issue comment: #377 (comment)

Handshakes weren't correctly passing a content-type header, making some browsers complain about errors parsing xml, although the data is really text/plain.

The pull req also contains two other minor style fixes as a bonus :P

… really make any sense

3rd-Eden · 2011-08-31T13:38:22Z

LTGM

Fixes issue commented on in #377

rauchg · 2011-08-31T15:52:08Z

lib/manager.js

@@ -775,6 +775,7 @@ Manager.prototype.handleHandshake = function (data, req, res) {
      res.writeHead(200, { 'Content-Type': 'application/javascript' });
      res.end('io.j[' + data.query.jsonp + '](new Error("' + message + '"));');
    } else {
+      res.writeHead(200, { 'Content-Type': 'text/plain' });
      res.writeHead(status);


Two writeHeads ?

Why 200 in writeErr? (the other 200 is for jsonp, which we have to do since it has no error handling)

The jsonp response originally had 200, so I kept it for the plaintext
response as well. This isn't as much an http error as a handshake
interpreted by script error, no?

On Aug 31, 2011 5:52 PM, "guille" <
reply@reply.github.com>
wrote:

@@ -775,6 +775,7 @@ Manager.prototype.handleHandshake = function (data,
req, res) {
res.writeHead(200, { 'Content-Type': 'application/javascript' });
res.end('io.j[' + data.query.jsonp + '](new Error%28"' + message +);
} else {

res.writeHead(200, { 'Content-Type': 'text/plain' });
res.writeHead(status);

Two writeHeads ?
Why 200 in writeErr?

Reply to this email directly or view it on GitHub:
https://github.com/LearnBoost/socket.io/pull/501/files#r108484

That's why the jsonp one passes an Error object so that we can do a quick instanceof check to know if the response was successful or not.
For the regular handshake, we of course just inspect the status code. That's why the original has res.writeHead(status), now duplicated.

Yeah, the duplicate writehead is bogus. And you're of course right about the
status code. I'll correct both.

On Aug 31, 2011 7:11 PM, "guille" <
reply@reply.github.com>
wrote:

@@ -775,6 +775,7 @@ Manager.prototype.handleHandshake = function (data,
req, res) {
res.w...
That's why the jsonp one passes an Error object so that we can do a quick
instanceof check to know if the response was successful or not.
For the regular handshake, we of course just inspect the status code. That's
why the original has res.writeHead(status), now duplicated.

Reply to this email directly or view it on GitHub:
https://github.com/LearnBoost/socket.io/pull/501/files#r108609

pyrostrex · 2011-08-31T22:15:46Z

@guille, @einaros, after some tests with socket.io after this update. It turns out that I must have einaros socket.io-client fork to get it work with FF6. Care to check it for me?. Thanks in advance!

I tried these :

Socket.IO (git cloned) and Einaros Socket.IO-client (git cloned) ---- works
Socket.IO (git cloned) and Socket.IO-client (git cloned) ---- fails
Einaros Socket.IO (git cloned) and Einaros Socket.IO-client (git cloned) ---- works

einaros · 2011-09-01T07:24:55Z

@pyrostrex, Use FireBug's network tab and verify that FF6 actually tries to use WebSocket as transport with the LearnBoost repo versions, and not actually falls back to FlashSocket. I've made a few updates to my socket.io-client, which allows even WebSocket to fall back to FlashSocket, so this may be what's happening with my builds.

pyrostrex · 2011-09-01T11:38:50Z

@einaros, I didn't enable flashsocket. I even use io.set('transports', ['websocket']). I installed firebug and see the network tab just like you did in your comment. and I can't see any "upgrade to websocket" request in firebug. Here's what i did to setup the socket.io:

cd node_modules
git clone https://github.com/LearnBoost/socket.io.git
cd socket.io
npm update
cd node_modules
rm -rf socket.io-client
git clone https://github.com/LearnBoost/socket.io-client.git
cd socket.io-client
npm update

But when I do the same, but instead of cloned from socket.io-client, I cloned from your repo. Then I refresh my browser. Then the firebug shows the "upgrade to websocket" request and everything else works.

pyrostrex · 2011-09-01T12:28:24Z

@einaros, It turns out I forgot to build. Now it's ok. :)

einaros · 2011-09-01T12:32:00Z

@pyrostrex, Ah -- well I'm glad you figured it out :)

To just straighten this out once and for all, though: you had to build the cloned LearnBoost/socket.io-client before you could get it working?

pyrostrex · 2011-09-01T13:23:58Z

@einaros, Yep! on my step. add another step which is "node builder.js". haha. Now trying to find a way to get socket.io websocket to work with latest windows chrome.

einaros · 2011-09-01T13:26:44Z

@pyrostrex, What you cloned from the socket.io repo should already have been
built, so that's unfortunate.

Regarding WebSockets with the latest Chrome - are you having trouble getting
that going? It should work out of the box. If it doesn't, which specific
version of Chrome are you running?

On Thu, Sep 1, 2011 at 3:24 PM, pyrostrex <
reply@reply.github.com>wrote:

@einaros, Yep! on my step. add another step which is "node builder.js".
haha. Now trying to find a way to get socket.io websocket to work with
latest windows chrome.

Reply to this email directly or view it on GitHub:
#501 (comment)

pyrostrex · 2011-09-01T13:41:26Z

@einaros, I'm using 13.0.782.218 m of chrome, like I said Windows only. In Mac, it doesn't have this problem. only in Windows. This time, it is much more weirder, it raises the "connection" event on the server but on the client, it still shows "pending" on the network tab.

einaros · 2011-09-01T13:50:12Z

@pyrostrex, Ok, 13.0.782.218 uses the old hixie-76 WebSocket standard, so it
isn't as such touched by the changes as of late.

Did you happen to see if my forks work with that version of Chrome?

On Thu, Sep 1, 2011 at 3:41 PM, pyrostrex <
reply@reply.github.com>wrote:

@einaros, I'm using 13.0.782.218 m of chrome, like I said Windows only. In
Mac, it doesn't have this problem. only in Windows. This time, it is much
more weirder, it raises the "connection" event on the server but on the
client, it still shows "pending" on the network tab.

Reply to this email directly or view it on GitHub:
#501 (comment)

pyrostrex · 2011-09-01T14:02:36Z

@einaros, It still happens with your forks. Here is my output:

Server output:

debug - served static /socket.io.js
debug - client authorized
info  - handshake authorized 10075011381301021817
debug - setting request GET /socket.io/1/websocket/10075011381301021817
debug - set heartbeat interval for client 10075011381301021817
debug - client authorized for

Browser Network Output :

URL    Method    Status    Type
/socket.io/1/?t=1314885489724    GET     200    text/plain
10075011381301021817/socket.io/1/websocket    GET    (Pending)    undefined

Additional Information :

Using Namespace.
Your Forks. Server and Client

einaros · 2011-09-01T14:04:09Z

@pyrostrex, Ok - does this work with the WebSocket transport on other
browsers than the Chrome13 you've tested with? If not, is your site behind a
proxy (such as nginx)?

On Thu, Sep 1, 2011 at 4:02 PM, pyrostrex <
reply@reply.github.com>wrote:

@einaros, It still happens with your forks. Here is my output:

Server output:

debug - served static /socket.io.js
debug - client authorized
info - handshake authorized 10075011381301021817
debug - setting request GET /socket.io/1/websocket/10075011381301021817
debug - set heartbeat interval for client 10075011381301021817
debug - client authorized for

Browser Network Output :

URL Method Status Type
/socket.io/1/?t=1314885489724 GET 200 text/plain
10075011381301021817/socket.io/1/websocket GET (Pending)
undefined

Additional Information :

Using Namespace.

Reply to this email directly or view it on GitHub:
#501 (comment)

pyrostrex · 2011-09-01T14:35:20Z

@einaros, I'm using pure nodejs.

Wow. It seems to also not working with firefox 6 Windows. For Mac, every browser works (chrome, firefox, safari). For Windows, everything doesn't work. I think it have something to do with the packet parsing thingy?

einaros · 2011-09-01T14:37:34Z

@pyrostrex, It's unlikely that it's because of the packet parsing, seeing as
this only affects hybi07->

See if http://draw.2x.io works from your Windows box. I've got a (fairly
recent) clone of my forks running there.

On Thu, Sep 1, 2011 at 4:35 PM, pyrostrex <
reply@reply.github.com>wrote:

@einaros, I'm using pure nodejs.

Wow. It seems to also not working with firefox 6 Windows. For Mac, every
browser works (chrome, firefox, safari). For Windows, everything doesn't
work. I think it have something to do with the packet parsing thingy?

Reply to this email directly or view it on GitHub:
#501 (comment)

pyrostrex · 2011-09-01T14:52:54Z

@einaros, I tried your website. It works but this is because of the websocket to flashsocket switching method. My windows is Windows Thin PC and I tried using chrome.

einaros · 2011-09-01T14:55:42Z

@pyrostrex, Chrome switched to FlashSocket for your connection to my site?
This is strange indeed - I'm running the exact same Chrome version you are,
and I get a WebSocket connection to it.

On Thu, Sep 1, 2011 at 4:52 PM, pyrostrex <
reply@reply.github.com>wrote:

@einaros, I tried your website. It works but this is because of the
websocket to flashsocket switching method. My windows is Windows Thin PC.

Reply to this email directly or view it on GitHub:
#501 (comment)

pyrostrex · 2011-09-01T15:01:02Z

@einaros, Yeah, maybe because my internet connection is extremely slow? 10kbps? I tried using Firefox 6.0.1, It never stop loading. haha.

einaros · 2011-09-01T15:02:40Z

@pyrostrex, Which addons are you using for Chrome and FireFox? And which
antivirus are you currently running? Does your antivirus come with any
additional firewall?

On Thu, Sep 1, 2011 at 5:01 PM, pyrostrex <
reply@reply.github.com>wrote:

Yeah, maybe because my internet connection is extremely slow? 10kbps? I
tried using Firefox 6.0.1, It never stop loading. haha.

Reply to this email directly or view it on GitHub:
#501 (comment)

pyrostrex · 2011-09-01T15:05:40Z

@einaros, Lol, I'm using fresh installed windows. Inside a VirtualBox. So no other stuff installed other than browsers.

Edit : Chrome and Firefox has firebug installed inside.

einaros · 2011-09-01T15:08:52Z

@pyrostrex, Hm, strange indeed. Your internet connection shouldn't be a case
at all for your local testing, and from what I gather you're really seeing
the same problems there, as you are against my site - save for the fact that
you've disabled flashsocket for your app. At this point you really should do
some digging deep with Fiddler or even WireShark, to see what is going on
with the WebSocket connections.

If you're able to do a clean WireShark dump of the WebSocket traffic from
your browser to your app, I'd be happy to look at that as well.

On Thu, Sep 1, 2011 at 5:05 PM, pyrostrex <
reply@reply.github.com>wrote:

@einaros, Lol, I'm using fresh installed windows. Inside a VirtualBox. So
no other stuff installed other than browsers.

Reply to this email directly or view it on GitHub:
#501 (comment)

pyrostrex · 2011-09-01T15:12:56Z

@einaros, Never know how to use WireShark. But I will try.

…syntax error message in the Web Console. Basically an addition to socketio/socket.io#501

einaros added 3 commits August 30, 2011 11:03

style changes

63fc15d

switched \u**** to \x** for single byte writes, as the former doesn't…

ae7f253

… really make any sense

added text/plain content-type to handshake responses

bd360a1

rauchg added a commit that referenced this pull request Aug 31, 2011

Merge pull request #501 from einaros/master

7c196f5

Fixes issue commented on in #377

rauchg merged commit 7c196f5 into socketio:master Aug 31, 2011

rauchg reviewed Aug 31, 2011
View reviewed changes

iamstarkdev mentioned this pull request May 17, 2022

[Snyk] Security upgrade engine.io from 1.4.0 to 3.1.0 iamstarkdev/socket.io#3

Open

geva mentioned this pull request Jan 9, 2023

[Snyk] Fix for 1 vulnerabilities geva/socket.io#32

Open

tjenkinson mentioned this pull request Jan 9, 2023

[Snyk] Fix for 1 vulnerabilities tjenkinson/socket.io#15

Open

abinabrahamanchery mentioned this pull request Jan 10, 2023

[Snyk] Fix for 1 vulnerabilities abinabrahamanchery/socket.io#8

Open

snyk-bot mentioned this pull request Jan 10, 2023

[Snyk] Fix for 1 vulnerabilities ebarahona/socket.io#8

Open

vanquisher0411 added a commit to vanquisher0411/socketio_project that referenced this pull request Feb 29, 2024

Firefox will try to parse the response from POST requests, causing a …

6fd91df

…syntax error message in the Web Console. Basically an addition to socketio/socket.io#501

darrachequesne added a commit that referenced this pull request Jul 8, 2024

[style] Add missing semi-colon (#501)

aa48d13

darrachequesne added a commit that referenced this pull request Jul 8, 2024

[chore] Bump uws to version 0.14.4 (#501)

519fb8d

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixes issue commented on in #377 #501

Fixes issue commented on in #377 #501

einaros commented Aug 31, 2011

3rd-Eden commented Aug 31, 2011

rauchg Aug 31, 2011

einaros Aug 31, 2011

rauchg Aug 31, 2011

einaros Aug 31, 2011

rauchg Aug 31, 2011

pyrostrex commented Aug 31, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

Fixes issue commented on in #377 #501

Fixes issue commented on in #377 #501

Conversation

einaros commented Aug 31, 2011

3rd-Eden commented Aug 31, 2011

rauchg Aug 31, 2011

Choose a reason for hiding this comment

einaros Aug 31, 2011

Choose a reason for hiding this comment

rauchg Aug 31, 2011

Choose a reason for hiding this comment

einaros Aug 31, 2011

Choose a reason for hiding this comment

rauchg Aug 31, 2011

Choose a reason for hiding this comment

pyrostrex commented Aug 31, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011

einaros commented Sep 1, 2011

pyrostrex commented Sep 1, 2011