I realized that the last time I heard Simon speak was at Seleimum conf in San Francisco in 2011. I've followed him on Twitter since then, so I feel I'm pretty well versed with what he's been up to, but the title intrigued me so much, I knew I had to be here.
Selenium has come a long way since I first set my hands on it back in 2007. During that time, I've become somewhat familiar with a few implementations and bringing it up in a variety of envirnments. I've reviewed several books on the tools and I've often wondered why I do what I do and if what I do with it makes any sense whatsoever.
Simon is explaining how a lot of environments are set up:
test <-> selenium server <-> grid <-> driver executable <-> browser
The model itself is reasonable but scaling it can be fraught with disappointment. More times than not, though, how we do it is often the reason it's fraught with disappointment. A few interesting tangents spawned here, but basically, I heard "Zalenium is a neat fork that works well with Docker" and I now know what I will be researching tonight after the Expo Reception when I get back to my evening accommodations.
Don't put your entire testing strategy in Selenium! Hmmm... I don't think we're quite that guilty, but I'll dare say we are close. Test the happy path. Test your application's actual implementation of its core workflows.
Avoid "Nero" testing: what's Nero testing? It's running EVERYTHING, ALL THE TIME. ALL THE TESTS ON ALL THE BROWSERS IN ALL THE CONFIGURATIONS! Simon says "stop it!" Yeah, I had to say that. Sorry, not sorry ;).,
Beware of grotty data setup: First of all, I haven't heard that word since George Harrison in "A Hard Day's Night" so I love this comment already, but basically it comes down to being verbose about your variables, having data that is relevant to your test, and keeping things generally clean. Need an admin user? Great, put it in your data store. DO NOT automate the UI to create an Admin user!
Part of me is laughing because it's funny but part of me is laughing because I recognize so many things Simon is talking about and how easy it is to fall into these traps. I'm a little ashamed, to be honest, but I'm also comforted in realizing I'm not alone ;).
Selenium has come a long way since I first set my hands on it back in 2007. During that time, I've become somewhat familiar with a few implementations and bringing it up in a variety of envirnments. I've reviewed several books on the tools and I've often wondered why I do what I do and if what I do with it makes any sense whatsoever.
Simon is explaining how a lot of environments are set up:
test <-> selenium server <-> grid <-> driver executable <-> browser
The model itself is reasonable but scaling it can be fraught with disappointment. More times than not, though, how we do it is often the reason it's fraught with disappointment. A few interesting tangents spawned here, but basically, I heard "Zalenium is a neat fork that works well with Docker" and I now know what I will be researching tonight after the Expo Reception when I get back to my evening accommodations.
Don't put your entire testing strategy in Selenium! Hmmm... I don't think we're quite that guilty, but I'll dare say we are close. Test the happy path. Test your application's actual implementation of its core workflows.
Avoid "Nero" testing: what's Nero testing? It's running EVERYTHING, ALL THE TIME. ALL THE TESTS ON ALL THE BROWSERS IN ALL THE CONFIGURATIONS! Simon says "stop it!" Yeah, I had to say that. Sorry, not sorry ;).,
Beware of grotty data setup: First of all, I haven't heard that word since George Harrison in "A Hard Day's Night" so I love this comment already, but basically it comes down to being verbose about your variables, having data that is relevant to your test, and keeping things generally clean. Need an admin user? Great, put it in your data store. DO NOT automate the UI to create an Admin user!
Part of me is laughing because it's funny but part of me is laughing because I recognize so many things Simon is talking about and how easy it is to fall into these traps. I'm a little ashamed, to be honest, but I'm also comforted in realizing I'm not alone ;).
No comments:
Post a Comment