I've got a large group of NUnit tests I have to import the outcomes from the given encounter a database, then characterize the group of results and offer these to the customers (email for test failures, web presentation for analyzing results). I have to be monitoring multiple runs with time, too (for confirming failure rates with time, etc.).
The XML would be the XML produced by nunit-console. I must import the XML having a minimum of fuss into some database that may then be employed to persist and offer results. We'll have many custom groups that people will have to have the ability to sort across, too.
Does anybody are conscious of a database schema that may handle posting this kind of data that may be personalized to the individual needs? This kind of problem appears like it ought to be common, and thus a typical solution should exist for this, however i can't appear to locate one. If anybody has implemented this type of solution before, advice could be appreciated too.
Obtain the build server to operate the tests, and it works of telling people what unsuccessful, and why.
I suggest TeamCity as it is several orders of magnitude simpler to setup.
I'm here searching to resolve exactly the same problem. We're presently leaning toward writing an XSLT to change the XML results into place claims. Then run the resulting file of place claims via a command line SQL interpreter. Ideally, I'd favour an NUnit add-in/extention that handles all of this for me personally. Regrettably I haven't been capable of finding one.
To construct off IainMH's answer you might want to have a look at using Trac with BITTEN, it's an free build system, that may run n-unit tests and report the outcomes. I presently utilize it for your exact functionality.
When utilizing MS SQL, you are able to import all XMLs to some common column of [xml] datatype. This, xpaths, searching and changes could be carried out.
Another Option to CruiseControl or TeamCity is Atlassians Bamboo if you are low on cash. I am a huge fan of the software for it's simplicity of use and they've an offer on where one can get bamboo for 10 dollars.
We have wished to avert this, but we have produced a database schema in the NUnit result XML schema it's a little deficient, however, because NUnit does some (inaccurate and strange) processing to find out a few of the critical statistics ("overlooked" versus. "not run", for instance).
We are still wishing to locate a schema / process that's Not really a complete CIT build system which could let us personalize a database for posting the outcomes, but presently we are utilizing a hands-folded database which we'll have to perform a large amount of designing onto obtain the preferred confirming.
Why must you possess the produces a database? Who's going for their services? The amount of failures can't be large. If it's (frequently) your development process is wrong. Fix the procedure. Eliminate waste (among the lean concepts), don't collect it.
Take more compact steps (shorter iterations, continuous build), eliminate dependencies.
This really is isn't generally done, because projects which have these type of problems don't deliver but get cancelled (eventually).
 Michael, monitoring nunit failures over a longer period provides zero value. You'll need a short feedback loop. Fix the issues now. Should you wait until you've gathered lots of problems, you will be overcome through the noise.
Good problem monitoring is performed in the right (greatest possible abstraction) level. Certainly not unit test.