Meanwhile, this one's for the webgeeks

[annathepiper]

The professional webgeeks, that is. As y'all know, I'm QA for the Seattle Times now and affiliated web sites. That means a whole heck of a lot of web presence. Right now we have no formal site regression procedures in place, and this really, really needs to change. I have a two-pronged goal here: one, set up what is essentially a site BVT, and two, set up a full-fledged, in-depth, bang the hell out of each and every functional aspect of the site functionality pass. These goals are going to have to come in stages, though. First we have to write the test cases, and then we have to figure out how many of them we can automate.

This is where you all come in. I'm looking for recommendations for web site testing tools, things that will do the grunt work of verifying links, check for missing graphics, that kind of thing--and ideally, also, let you focus the scope of what you're looking for so that (say) if you're testing all the links on seattletimes.com, it doesn't wander off into testing anything that's not part of that site. (We do have tools that are supposed to do this right now, but they're unreliable and frequently have to be double-checked manually. Not helpful.) For extra bonus points, an easy to understand UI, ability to save stuff out into readable log files, and ability to set up custom scripts/macros/actions without having to have huge gobs of coding skill would be bonus.

Commercially produced software would be fine, as would solid, reliable open source products. Hit me with your recommendations, people!

Comments

[xpioti.livejournal.com]

WebTrends has been wildly popular with my previous clientele. I have no idea what my current clientele uses, assuming they use anything at all. :)

[annathepiper]

Hrmm, unless there's information that I'm not seeing on WebTrends' web page, though, this doesn't quite seem like what I'm looking for. It seems like more of a package to measure site metrics. But what I'm looking for is more testing the actual functionality of the site.

[fleetfootmike.livejournal.com]

JMeter (open source) doesn't suck.

You can use the GUI to set up a bunch of tests, then run them in batch mode by the command line once you're happy.

[annathepiper]

Cool, thank you!

[kirbyk.livejournal.com]

One solution, a very DIY approach, would be to write custom Perl to do it. There's a great module, WWW::Mechanize, that lets you define a path of URLs, handles cookies, and that sort of thing. It's probably the most work to take this approach, but you can't beat it for the range of what you can do. It might be a good kind of thing to fill in the gaps if you find an easy tool that does 90% of what you want, too. Or if you want to do more programming.

http://search.cpan.org/dist/WWW-Mechanize/

Article on using it: http://www.perl.com/pub/a/2003/01/22/mechanize.html

[annathepiper]

The problem with this approach, though, is that for me at least it would require huge gobs of coding skill. I've dabbled with Perl maybe once or twice, max. I can code, but I can't code fast, and the time investment in learning how to set up a Perl script to make it do what I want to do would be huge. Probably to the detriment of actually manually testing what needs to get tested. And we've got a temp right now, but I'm not getting the impression from him that he's strong enough on coding that he could whip up test scripts with ease.

We have days where we could effectively have one of us spend time on coding and the other on testing what needs to get looked at, but only occasionally. More often we have maybe more like 1 1/2 to 2 people's worth of work that needs immediate attention.

This might be a nice long-term thing to fill in gaps, though, as you say, once we find a tool to solve most of the needs. So thanks for chiming in. :)