Hi, I downloaded SeeShell to evaluate its feasability for a bigger automation.
It’s concept and approach sparked our interest but unfortunately we need to conclude that SeeShell is all but “rock solid”.
test scripts arbitrarily stop their execution - they run endlessly. It can be stopped from main window then
Simple operations like “Type” (aka Send Keys) does not work reliably:
e.g. command Type “{WINKEY}chrome https://abcdef…{ENTER}”. Only works 1 out of 10 cases.
Either only the start menue opens, sometimes chrome starts but only has a fraction of the URL line in the address bar. ~1 out of 20 cases it is working fully.
=> It looks like a timing issue with the sendkey buffer to us. That is really basic stuff…
EXTRACT does not work reliably at all.
E.g error message “no green rectangle found”-> But it is definetely there in the image;
Testing the recognition with the image editor shows very ( I mean really very) odd matches all around the screen. These matches have not even the same color as the pattern. Also increasing confidence level does not solve it…
The built in chromium browser dies. It usually works once immediately after startup. Then after the first script running chromium dies.
As this is a bunch of not obviously related but very fundamental issues I am wondering if there is a stable version available for us to test where at least basics would work to evaluate?
Hello, thanks for the open feedback. This is a long list of issues, so it seems something is going wrong and SeeShell is not working as it usually does: So I have a few more questions:
Can you please describe what kind of tasks you want to automate?
Do you use SeeShell for Web Automation, SeeShell for Desktop automation or both?
Hello,
many thanks for the very fast reply. Sure thing, here are more informations:
What kind of tasks we intend to automate: -> Transferring of (ticket) data from one web application to another web application
We have tried both, “Desktop” first as it sounded more promising (can do both), then SeeShell for Web. The latter works slightly better admitted (application has only died twice until now and recocnition seems more successful)
Thanks for the info. Actually… for this use case I recommend that you use our Kantu software. Reasons:
With storeText, storeValue and sourceExtract Kantu is better at extracting info from websites (web scraping)
In addition to selenium ide commands, it supports now the same visual automation as SeeShell, including relative clicks. (However, it has no recording wizard for image commands. The visual commands are created within the Kantu editor.)
One drawback is that it does not have a COM API. But it has a command line interface, so it can be scripted as well.
Thank you very much for giving this (workaround) recommendation.
Just experimenting with Kantu… unfortunately first tests show more problems, this time with XClick (required as “click” does not work/fire up the js event on the web app:
-> “kantu-xy-host.exe has stopped working”; will open seperate thread under kantu