Mozilla Firefox addon error - Ranorex 7.2.0

[odklizec]

Hi Ali,

Could please share a Ranorex snapshot of the web page in question and some xpaths, pointing to the problematic elements? Also, a sample page (or HTML code) is pretty critical here. Without, at very least, a small sample, it would be next to impossible to debug and fix the problem.

As a new forum member, you may not be able to upload attachments, but you can still use a 3rd party file sharing service, like onedrive or dropbox, or eventually send the samples directly to Ranorex support to [email protected]. Thanks.

[Aracknid]

Has this been fixed in Ranorex 8?

[ahoisl]

Has this been fixed in Ranorex 8?

Depends on what you mean by "this".
The console popping up when clicking the Ranorex icon in Firefox is fixed. We fixed quite some bugs on the firefox addon, however, we did not change anything specific to performance with iframe. Sorry...

We have analyzed the speed problem and it seems that the new Firefox extension system is just slower for some operations, especially with large websites having a lot of iframes. We were not able to find a workaround, yet...

Best Regards,
Alex
Ranorex Team

[odklizec]

Hi Alex,

I think he means the overall slowdown of tests in FF with new plugin. It's not related to just iframes. I made a small sample solution (already provided to support - Case-00030832), which does simple enumeration of links (atags) at Ranorex page and it shows noticeably slow operation in FF.

PluginSpeedComp.png (20.16 KiB) Viewed 538 times

Here is the code I'm using in sample solution:

Code: Select all

        public void EnumLinksAndMoveMouse(RepoItemInfo linkInfoElements)
        {
        	// Create a list of adapters using the "Info" object
			IList<ATag> atagList = linkInfoElements.CreateAdapters<ATag>();
			// Move the mouse pointer to each link of the list
			foreach (ATag atagElement in atagList )  
			{  
			    atagElement.MoveTo();
			    Delay.Duration(100);
			}
        }

It seems that the links enumeration is somewhat faster with Ranorex8 plugin (comparing with Ranorex 7.x), but still noticeably slower, than in IE and Chrome.

[Aracknid]

Yes, I meant the overall slowdown on FF. Hopefully you find a solution soon.

Aracknid

[odklizec]

Hi all,

I would like to refresh this post a bit. I just tried my FF slowdown-check test with Ranorex 8.0.1 and it seems the speed of FF plugin has not changed much in this release? It's not very surprising, since there was nothing mentioned in 8.0.1 release notes. The slowdown shown in this particular test is not too obvious, but in my production test, the numbers are much more significant. It makes a difference if the test, which takes less than 20m in Chrome, takes over 40m in FF

IE_FF_Chr.png (20 KiB) Viewed 433 times

Additionally, FF plugins often throws below exception at this lines:

Code: Select all

IList<ATag> atagList = linkInfoElements.CreateAdapters<ATag>();  

Excpetion:

Code: Select all

LineageGlueRule 'firefox_web_child_elements' caused RanorexException on Execute. Error in MsgHostDom call: Error TypeError: doc is null @moz-extension://d13d92ab-387d-43cf-b731-e43d4c581ae9/content.js:81:1 [email protected]://d13d92ab-387d-43cf-b731-e43d4c581ae9/msgport.js:11:13 
Show/Hide Stacktrace
Ranorex.RanorexException: Error in MsgHostDom call: Error TypeError: doc is null @moz-extension://d13d92ab-387d-43cf-b731-e43d4c581ae9/content.js:81:1 [email protected]://d13d92ab-387d-43cf-b731-e43d4c581ae9/msgport.js:11:13 System.Runtime.Remoting.Proxies.RealProxy.HandleReturnMessage(IMessage reqMsg,IMessage retMsg) +0x27 System.Runtime.Remoting.Proxies.RealProxy.PrivateInvoke(MessageData& msgData,Int32 type) +0x1b3 Ranorex.Plugin.FirefoxIpc.IFirefoxDom.GetFrameElementChildren(NpPtr elemId) +0xffffffff Ranorex.Plugin.FirefoxIpc.FirefoxDom.GetFrameElementChildren(NpPtr elemId) +0x0 Ranorex.Plugin.FirefoxWebChildrenRule.Execute(GlueRuleExecutionState state) +0x109 Ranorex.Core.ElementEngine.ApplyGlueRules(Element element,GlueRuleExecutionStateImpl state,Boolean includeRemotes) +0x6a 

[ahoisl]

Actually, we are currently working on improving the performance of Firefox automation. Unfortunately, Firefox is slower than Chrome, no matter what we do, but we will try to significantly improve whatever we can from our side.

I cannot promise you more at the moment, as we are still in an experimental stage. But I'll keep you posted.

Regards,
Alex
Ranorex Team

[odklizec]

Hi Alex,

Thanks for the heads up! Of course, I don't expect the same performance from each browser, but any improvement in this area is welcome There is already a noticeable improvement between 7.x and 8.x. I think the speed issue is partially caused by the mentioned exceptions? I see them quite a lot in reports. Feel free to contact me in case you need more details or to test something.

[ahoisl]

If you have not already, make sure to check out Ranorex 8.1.0 which boosts performance with Firefox
After installation, please also update the Ranorex Firefox browser addon, thank you!

Regards,
Alex
Ranorex Team

[odklizec]

Hi Alex,

I already did Unfortunately, there appears to be a new problem with FF plugin, which somewhat nullifies the speed improvement in 8.1. In some test cases, there is generated a large number of LineageGlueRule errors, which seems to cause another speed degradation. It's already under investigation...
https://www.ranorex.com/forum/lineagegl ... 12017.html

[odklizec]

I'm afraid, I was too quick about sharing a good news, regarding the speed improvement in FF plugin, I posted earlier today in another forum thread. The speed was definitely improved in 8.1/8.1.1 (over 7.x). However, some of my test cases are still much slower than the same tests running in Chrome. I'm trying to nail the problem down to certain code-based actions (it appears to be related to CreateAdapter and GetAttributeValueText). So far, I'm unlucky to reproduce the problem outside our jqwidgets-based app, which I can't share. So I'm going to try to simulate the problem with jqwidgets demo page. I will keep you informed.

[odklizec]

OK, here is a quick&dirty sample, which shows the speed difference between IE, Chrome and FF. The code used in this sample is most probably not the only cause of the slowdown in FF. I will try to identify other operations, which I'm using in my real-world tests and which are causing overall test slowdown (in FF).

The speed difference between Chrome and FF may not be so obvious in this sample, but in my real-world test (running on VM Windows Server 2012 R2) I can see the difference like 8m runtime in Chrome and 12-15m in FF. What's interesting, IE is fastest of all three browsers.

[Support Team]

Hello odklizec,

I would like to inform you that our developers have created an issue in our TFS system and will investigate based on the sample that you have provided. Unfortunately, I am currently not able to say when this investigation will be done.

As soon as I have some information regarding the issue, I will let you know.

Sincerely,
Tomaž

[odklizec]

Hi Tomaz,

Thanks for the heads-up. I can live with somewhat slower FF tests for now. Let's hope your devs will find a way around