CI is python scripts, maybe adaptable without too much work?
Jacques
From: "Scott Gray" <scott.g...@hotwaxmedia.com>
That sounds like a good idea to me. What would be awesome is if the
CI server could inspect the commit, determine the components/
applications affected and then only run the applicable tests. Full
tests runs could be reserved for framework commits or something like
that.
Regards
Scott
On 16/12/2009, at 3:13 AM, Tim Ruppert wrote:
One thing that we could easily do is put together a continuous integration that has several branches - one that is run
immediately upon each of the commits and one that is scheduled for a few times a day for these longer running tests. I agree
that we definitely need these in place, but breaking them up might:
1. Get people to write more unit tests that are quicker to run.
2. Get people to write more of the longer ones because those will still be
catching any potential issues.
Just a thought.
Cheers,
Ruppert
--
Tim Ruppert
HotWax Media
http://www.hotwaxmedia.com
o:801.649.6594
f:801.649.6595
On Dec 15, 2009, at 4:01 AM, Matthieu Bollot wrote:
Le mardi 15 décembre 2009 à 23:28 +1300, Scott Gray a écrit :
Hi Erwan,
It'll be another couple of days before I can make an informed comment,
I'm still very much in the learning phase.
One question I do have, does anybody have selenium setup to run in a
continuous integration environment?
We do have some at nereide
If you have lots of tests, how long to they take to run? Is
SeleniumGrid a good solution to shortening the time a test run takes?
So maybe a few questions :-)
The results here :
http://selenium.neogia.org/ofbiz/results/result_2009.12.14_08:12:44.html
took 12 minutes for 13 tests
and the results here :
http://selenium.neogia.org/ofbizNeogia.stable/results/result_2009.12.14_08:12:45.html
took about 20 minutes for 41 tests
That's a quite fast server. 4 proc, 8GB ram (those are html testSuite
using selenium-server.jar directly, perhaps using seleniumXML can change
the duration a little bit)
And loading a page is may be the only "long" stuff.
I guess I'm still sitting here wondering if WebTest isn't a better
solution simply because it doesn't require a browser (just another ant
task) and the tests run faster. I've never used either before so I'm
in the dark on these solutions.
I think that selenium is quite good because you can _really_ know how
long took a test, e.g we could also test how long it takes to purchase
1000 products (don't know if it will be still possible with webtest)
cheers,
Regards
Scott
HotWax Media
http://www.hotwaxmedia.com
On 15/12/2009, at 10:07 PM, Erwan de FERRIERES wrote:
Hi all,
As many of us are now looking into seleniumXML, I would like to
discuss a bit more with you of the logging of errors and success in
seleniumXml.
Has anyone started something ? The changes that have to integrate
are major and is would be great to coordinate our efforts.
What I'm thinking is adding JUnit asserts at the end of a selenium
command, to be able to create JUnit XML files and after creating a
report. This will then help us to identify errors on the interface
or in functional testcases.
Regards,
--
Erwan de FERRIERES
www.nereide.biz
--
Matthieu BOLLOT
www.nereide.biz
