Re: nunit improvement

Hi,
 
I'm copying this to the user list, since we are getting into some things of
general interst.
That's actually the best place to ask questions because (1) other people can
answer (I 
might be on vacation and (2) other people can benefit.
 
You asked why the tests were green even though the fixture is red - Well,
the tests
all ran successfully and passed individually. The error occured in
TestFixtureTearDown,
which is not part of any of the test cases.

TestFixtureTearDown is not considered part of the test. The idea is that
TestFixtureTearDown
should be able to successfully run in all cases. If you look closely, you
will see that 
your fixture shows an error, not a failure. The difference is subtle, but
see how the
message shows an unexpected exception (NUnit.Framework.AssertionException).
In
normal failures, only the message is shown.
 
You can continue to do this, so long as you understand how to interpret the
result. In
the next release, I'm trying to show such errors more clearly. Up to now,
all releases
of NUnit have only counted test cases errors in reporting results.
 
Alternatively, you may be able to put your check in a TearDown method, which
is
repeated for each test and results in the individual test failing, or in the
test methods
themselves. Of course, this means you would need to setup and delete your
application
for each test and that may not be what you want to do.
 
Regarding your suggestion that old results in the tree be kept across runs,
it makes
sense to me. Would you put it in as a feature request on SourceForge,
please?

Charlie

  _____  

From: PIN Yannick [mailto:[email protected]] 
Sent: Tuesday, April 29, 2008 2:21 AM
To: Charlie Poole
Subject: RE: nunit improvement



Hello,

Thank you again.

I’m sorry to disturb you again. I hope that it is the last time :-)

 

In fact you are all right.

I verify the test and there is an assertion inside the
[TestFixtureTearDown]:

Assert.AreEqual(oldFree, newFree);

 

at the end of the test, I want to delete application and I want to verify
memory recovery.

The way that is used today is in the testfixtureteardown (delete appli,
logoff, and verify memory).

Is there another way to generate an error in case of bad memory recovery
after the deletion, than use an assertion that seems to cause problem ?

 

Moreover, I have another question. Is it possible to click on the button
”run” without deleting the status of test already run (color), and deleting
only the status of tests that will be run ? I try several options (uncheck
the option “clear results when reloading, activate “save visual state of
each project” (but it is not convenient for performance purpose)), but the
colors are always lost.

The purpose is to be able to avoid to run the whole campaign in one step but
run the whole campaign or a subset in several time and being able to save
the status obtained in xml file.

I don’t manage too see in the source code where the clean of the color of
the treeview is done.

 

Thank a lot.  Regards.

 

PIN Yannick 
Validation team
Gemalto 
Tel: +33 4 42 36 30 62
Fax: +33 4 42 36 58 09
Avenue du Jujubier, ZI Athélia IV – BP90
13702 La CIOTAT
[email protected]
 <http://www.gemalto.com> www.gemalto.com
  


  _____  


From: Charlie Poole [mailto:[email protected]] 
Sent: lundi 28 avril 2008 19:47
To: PIN Yannick
Subject: RE: nunit improvement

 

Hi Pin,

 

Glad that helped.

 

Rearding the color, I meant to ask about the properties of the suite that is
red. However, the message behind the properties display seems to give the
answer. You appear to have placed an Assert in the TestFixtureTearDown
method of TestGSCIS. This causes an error in the teardown, after all the
tests have run. It's generally not useful to put an Assert in
TestFixtureSetUp or TestFixtureTearDown fjor this reason.

 

Charlie

 


  _____  


From: PIN Yannick [mailto:[email protected]] 
Sent: Monday, April 28, 2008 8:59 AM
To: Charlie Poole
Subject: RE: nunit improvement

Hello,

Thank you very much for your fast response.

According to your answer in fact, I have found the corresponding option.

It solve my issue for the loading of the campaign.

 

For the properties of the test, see attached file.

In fact, I though it was a color bug of the tools but I notice a wrong
comparison behind (errors and failure tab)

Perhaps it is the reason but I do not understand why the test himself is not
red.

 

Thank you very much for your support.

 

 

Regards.

 

PIN Yannick 
Validation team
Gemalto 
Tel: +33 4 42 36 30 62
Fax: +33 4 42 36 58 09
Avenue du Jujubier, ZI Athélia IV – BP90
13702 La CIOTAT
[email protected]
 <http://www.gemalto.com> www.gemalto.com
  


  _____  


From: Charlie Poole [mailto:[email protected]] 
Sent: lundi 28 avril 2008 16:50
To: PIN Yannick
Subject: RE: nunit improvement

 

Hi Yannick,

 

I'll take a look at how that can be made more efficient but you can always
turn it off in the options.

 

Regarding the red suite with green tests, this is actually valid in some
circumstances. What do

the properties for that suite indicate?

 

Charlie

 


  _____  


From: PIN Yannick [mailto:[email protected]] 
Sent: Monday, April 28, 2008 3:06 AM
To: [email protected]
Subject: nunit improvement

Hello,

I have a performance issue with nunit 2.4.7.

I have a big campaign.

In fact, when you start nunit, it read a file (x_visual_state.xml) that is
very large (3 Mo) and because of that, the loading duration is about 15
minutes.

When I modify the file TestSuitetreeview.cs by adding a Boolean firstLoad to
avoid the first load, the performance is ok (about 20s to start NUnit).

Perhaps an improvement can be done in the xml file (not store all campaign
status but only open item) ?

Moreover, I have a color bug. All tests are green in a directory, but the
parent directory is red.

The same problem exists in 2.0 version so it is an very old one.

 

<<TestSuiteTreeView.cs>> 

But in fact, it is a good tool. Congratulation for it :-):-)

I hope that my remarks can improve the tool.

Regards.

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone

_______________________________________________
Nunit-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/nunit-users