cancel
Showing results for 
Search instead for 
Did you mean: 

Disappearing Spy Modes

GAVIN.RUDLING
Level 7
Good Day All,

I'm wondering if any body is also experiencing spy modes that aren't available.
I am using Blue Prism Versions 6.10 and 7 and when I am using the Chrome Application model. I seem to loose all the spy modes except for Browser Mode.
Funny enough the first time I use it it's fine. I have Win32 and UI methods, but the next time I try add on some elements they don't seem to be available. This I also find on multiple machines. Does anyone know of a solution?

Thank you kindly
Gavin

------------------------------
Gavin Rudling
Digital Consultant & Developer
Cog3nt
Africa/Johannesburg
0813906789
------------------------------
1 BEST ANSWER

Helpful Answers

I know the below may sound somewhat counter-intuitive, but this is based on the assumption you only have one instance of Chrome open.

In your application modeler, try removing the window title and have it just attach to an existing instance of Chrome. In your attach stage, try leaving the Window Title and Child Index fields blank as well. Child Index unfortunately seems to be unreliable due to a somewhat random load order within Chrome; Window Title may still end up attaching to an incorrect sub-index with the same identifying name.

For whatever reason, I've consistently seen where leaving this data out results in attaching to the correct instance. This is of course problematic if you're trying to leverage multiple instances simultaneously, but there are probably ways of getting that to work fine.

Edit: After re-reading your post a few hours later, there's also a sort of hacky workaround that may get you what you need. Since you're already using the --force-renderer-accessibility flag, try creating a new object with an application model set to an existing Windows application (specifically do NOT use a browser model). This'll allow you to attach to Chrome and use both Win32 and Active Accessibility modes. (I personally find AA to be better than UIA anyways.)
------------------------------
Ami Barrett
RPA Blue Prism SME
Karsun Solutions
Plano TX
------------------------------

View answer in original post

7 REPLIES 7

KodiSrihari
Level 8
Hi, 

    As i also face some how like the same that , In chrome application the spy modes can't able to spy correctly. But after changing the Screen Resolution Scale Settings to    100 % the Spy modes Worked Perfectly.

Thanks.

------------------------------
Srihari
RPA Developer
------------------------------

Hi and thanks Srihari.

 

These are my current settings. Are there any other settings I should look at?

 

Regards

Gavin

 

21698.png

 



Hi Gavin, 

It seems settings Good as per my knowledge. 

Thanks

------------------------------
Srihari
RPA Developer
------------------------------

AmiBarrett
Level 12
Gavin,

I have previously seen where Automate is attaching to the wrong instance of Chrome and shows this behaviour, particularly in an object trying to attach to an existing instance. Is this object the same one which launches your browser? If not, can you send a screenshot of your attach inputs?

------------------------------
Ami Barrett
RPA Blue Prism SME
Karsun Solutions
Plano TX
------------------------------

Hi Ami, Thank you very much for your response.

 

I did think it's was a attaching issue at a stage.

I found that in my one object which seems to be working, I am attaching to a Portal page with the Title of "Claims"

The next object is trying to attach to kind of popup screen an this page title is "Claim" without the "s".

When I am trying to attach the error throws: Unfortunately Blue Prism automatically places the (*) at the end, so I will still pick up the "Claims" title.

 

21712.png

 

21713.png

 

I have a Launch object, Login object, and an Actions object.

Launch and Login don't seem to have any issues. Launch obviously launching chrome and Login object attaching.to the launched instance.

The Actions object also attaches to the launched instance, but has issues. The spy methods were there at initially because I did spy the elements at a stage using the Win32 and even UI.

For some reason they disappear after a while and I don't know why. I may be over looking and not seeing at this stage.

Earlier today I thought of trying to build a new object, but still no joy. I renamed the first object and somehow the spy methods returned and I was able to continue.

But again after a while when stepping through the process, I got errors. I checked the elements and they all failed No Matching  elements found. These are all the elements that were spied using Win32 and UI. elements using the Browser mode are fine.

 

Thank you kindly again

 

Regards

Gavin

PS: Just realised that the ones that are working are working because all the elements were spied in Browser mode. However the bottom left image does show that elements were spied using "UI",

But because I used two spy methods in the wait stage, which ever one is picked up first will be used. So Browser mode is obviously being picked up as UI mode is no longer available.

A bit of a fail safe if I have multiple options to spy the elements.

 

                                                                      Working                                                                                                                                                                       Not Working

 21714.png                                           21715.png   

21716.png                                            21717.png

 

 



I know the below may sound somewhat counter-intuitive, but this is based on the assumption you only have one instance of Chrome open.

In your application modeler, try removing the window title and have it just attach to an existing instance of Chrome. In your attach stage, try leaving the Window Title and Child Index fields blank as well. Child Index unfortunately seems to be unreliable due to a somewhat random load order within Chrome; Window Title may still end up attaching to an incorrect sub-index with the same identifying name.

For whatever reason, I've consistently seen where leaving this data out results in attaching to the correct instance. This is of course problematic if you're trying to leverage multiple instances simultaneously, but there are probably ways of getting that to work fine.

Edit: After re-reading your post a few hours later, there's also a sort of hacky workaround that may get you what you need. Since you're already using the --force-renderer-accessibility flag, try creating a new object with an application model set to an existing Windows application (specifically do NOT use a browser model). This'll allow you to attach to Chrome and use both Win32 and Active Accessibility modes. (I personally find AA to be better than UIA anyways.)
------------------------------
Ami Barrett
RPA Blue Prism SME
Karsun Solutions
Plano TX
------------------------------

Hi Ami, thanks again for your time and effort. I think I have resolved the issue using some of your suggestions.

 

For some reason I could not make the changes on the existing Object. It kinda had no effect.

 

What I did was create a new object and instead of using the browser already running option, I used the launch from executable for all the objects  and just did not provide the URL details.

 

21726.png

 

Then each action attaches to the process name with a child index of 0.

This somehow enabled the other spy modes other than browser mode.

 

Thanks again Team, much appreciated.

 

Regards

Gavin