Hi,

sorry but i can not reproduce this issue at least at my configuration: Vista+IE9+32b.

Could you repeta the problem with IE driver debug log?
Here is an example of setup such log https://gist.github.com/3176482
Maybe it helps to understand why it's happened.

Hello,

Thanks. Here's the log. Well, line 396 is suspicious (Unable to add cookie, document
does not appear to be an HTML page)

I played with it a bit and here's what I figured out. There may be problems with parsing
cookie files. I've set a breakpoint in the second test and looked at getCookies() —
there was one item. IE's devtools report there to be two cookie items. In the project
I'm actually testing getCookies() also return a list of two or three items, whereas
six are actually set.

Hi,

thanks. i think it'll help. Major lines, imho, are:
-------------------------------------------------------------------
T 2012-07-25 18:06:20:189 DocumentHost.cpp(253) Entering DocumentHost::AddCookie
...
T 2012-07-25 18:06:20:198 DocumentHost.cpp(310) Entering DocumentHost::IsHtmlPage
...
D 2012-07-25 18:06:20:199 DocumentHost.cpp(329) Unable to support custom document type:
Opera.Protocol
...
W 2012-07-25 18:06:20:200 BrowserFactory.cpp(589) RegQueryValueEx failed for retrieving
required buffer size
W 2012-07-25 18:06:20:200 DocumentHost.cpp(359) Unable to read mime type from registry
for document type
W 2012-07-25 18:06:20:200 DocumentHost.cpp(266) Unable to add cookie, document does
not appear to be an HTML page
-------------------------------------------------------------------

For it looks like Opera is installed or was be installed, maybe as default browser.
If IE is not "default", it's tried to check content of HKEY_CLASSES_ROOT/.htm element.
Usually there is string, i.e. <XXX> which can be used to get content of another HKEY_CLASSES_ROOT/<XXX>
element
which contains something like document type, i.e. for me they are "FirefoxHTML" and
"Firefox HTML Document".
From last got value we try to determine is it HTML page (opened by HTML processing
program) or not.

Found error said about it can not find HKEY_CLASSES_ROOT/Opera.Protocol, don't know
why.

Could you describe which browseres are installed at your machine, which is default?
Looks like yet another trick for Opera is required at DocumentHost::isHTMLPage().

But, it's for someone who will fix it, i really don't understand for which purpose
it is still required? It has trick for IE, FF and looks like will for Opera.

I hope it helps.

Thanks again. Yes, Opera is my default browser. It's weird, imo, that it causes such
error.

The following browsers are installed on my system:
• Internet Explorer 9 (32- and 64-bit);
• Opera 12.0 (default);
• Mozilla Firefox 10.02 (started updating right now :) );
• Google Chrome 20.0.1132.57
  and Google Chrome Canary 22.0.1217.0;

Concerning the tricks you mentioned. Does that mean that if I want to test for IE,
I need to keep it (or FF) a default browser? If that's a solution, well, humbly I think
this should be fixed so that running any webdriver does not depend on default browser
(document type *description*).

Can't it be checked by mimetype (text/html)?

Thanks for information.

No, you don't need to setup tested browser as default. I just described how it works
at selenium level. Potentially all browsers configuration should be supported, imho.
Just for info - could you check workaround - settin gup tested IE as default and run
the same test - is it helped? 

Yes, it can be checked in other way at selenium level.
I think, someone from project members will fix it. I'm only helping with investigation
of issue reason and getting additional information.

Okay. Yes, really, with either IE or Firefox or Chrome set up as default, the test runs
successfully. Having Opera set up as default makes the test fail.

Thank you for your help a.u.savchuk. Do I need to take any further steps so that Selenium
devs see this issue, or they just gonna visit this topic eventually? (I'm first time
to issue trackers, you know *blush*)

Could you re-run your test with the latest private release of the IEDriverServer.exe?
I've updated the logging for getting values from the registry which should help diagnose
this a little further. You can get the latest 32-bit private release from http://selenium.googlecode.com/svn/trunk/cpp/prebuilt/Win32/Release.

Done. However I diffed the logs — no much difference. Opera is set up as default again.

If you need registry values alone, I can provide them. In fact, they're pretty much
the same as if FF was default, except the word Opera instead of Firefox.

You should have seen additional logging around the registry retrieval. Did you download
from the above link, or did you go through either the SeleniumHQ website or http://code.google.com/p/selenium/downloads/list?
The additional logging hasn't been pushed to be publicly available yet; you'll have
to get the prebuilt server directly from the source tree in Subversion. The link in
comment 9 is a web view of that repository.

I have downloaded IEDriverServer.exe from the index you linked to, replaced my old IEDriverServer.exe
with this one and rerun the tests. Then attached the log. Do I need to do any other
actions except that? Just in case I have to rebuild something from source, I fear I
won't be able to do this. I'm a complete newborn.

(however, I'm not that stupid :) )

This issue was closed by revision r17591.

I am having this issue with selenium node 2.33, iedriver 2.35.  Attached is the IEDriver
log.  Was this re-broken?

@ricardo.j.lopez 

Please raise new issue with:
* full executable test script
* IE browser information
* OS information
* IEDriverServer.log with TRACE level

I've been seeing the issue as well - Ie10, Windows 7, IedriverServer  64-bit (dec 5th
2013 version) and selenium standalone server 2.38.0.. 

For the people at our office here, it appears that they can run the tests in their
local IE instances, but if they attempt to run using sauceLabs or remoteWebDriver,
the cookies don't get set at all.

We've tried adding cookies the "usual" route and also thru javascript - but neither
seem to do the trick.

I have the same problem too. Unable to set cookie. Is there a fix?

ie9 32bit. Selenium 2.37. iedriver 2.38.

Please see the attached file. Thanks!

I'm afraid the logs attached to the comments do not match what the comments say are
being run. The log in comment 18 shows running 2.31.0.0 of IEDriverServer.exe; the
log attached to comment 19 shows running 2.28.0.4. The fix for this issue wasn't applied
until 2.37.0.1, in revision 334d39caf93. Please retest with at least that version,
and re-post your logs.

I'm unable to get any cookies it seems with IE.  The same code works fine with FF. 
Just a simple getDriver.manage.getCookies returns nothing in IE.  using 2.35

execute javascript for setting cookies with protective mode on.

Hope we could fix it without having to do some workaround in JS just due to an HTTP
header 'cache-control: no-cache' with IE9, selenium server 2.42.2 and IEdriver 2.42.0.
See this issue which is maybe related :
https://code.google.com/p/selenium/issues/detail?id=6985#c9

Thanks