Gary, 

I appreciate your input and I'm trying to digest it now.  How do you
reference the private methods you will be testing if they are in a
separate class (or project)?  

Secondly, is this separate unit testing class in the same solution?  Or
another solution altogether.

Lastly, I've only seen examples of conditional compilation using only
Debug or Release.  If I decide to go with my original route with a
separate UnitTest configuration (which mirrors Debug in every way with
the addition of a new preprocessor directive called TEST), how do I set
this build up in a .build file?

Thanks,

Brian

> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:nant-users-
> [EMAIL PROTECTED] On Behalf Of Gary Feldman
> Sent: Tuesday, March 30, 2004 8:03 AM
> To: [EMAIL PROTECTED]
> Subject: Re: [Nant-users] Nant & Unit Tests From Conditional
Compilation
> 
> ----- Original Message -----
> From: "Brian Beaudet" <[EMAIL PROTECTED]>
> 
> > What I'd like to do is make a build and run the unit tests (as a
build
> > verification test).  Now, should I only build the UnitTest
configuration
> > I've defined and is that a valid approach?  Or should I run the
UnitTest
> > configuration, run the unit tests and if they all pass, then run my
> > debug or release configuration (depending on my needs at the time)?
> 
> You need to identify what you're trying to accomplish, and what the
> resource
> costs are.
> 
> In the typical nightly build situation, the primary purpose is to
identify
> any newly introduced bugs.  The product of the build won't be going
> anywhere.  Thus the usually approach is to do a debug  build, and then
run
> the tests on that.  If anything shows up, then the developer have a
ready
> build to debug (which is a particular advantage if their development
tree
> doesn't match the nightly build).  Conversely, because developers are
the
> only ones likely to see a debugging build, there's no harm in
including
> the
> extra testing code in the output.
> 
> Release builds are trickier.  If your test code is small and
unobtrusive,
> you may not care if you ship it to the end users.  But often that's
not
> acceptable.  The bottom line in that situation is that you either have
to
> do
> two builds, or forego running the unit tests in the Release build.
Either
> way, you're not going to be unit testing the final product - which may
or
> may not be ok, depending upon the quality of the rest of your test
system,
> as well as your bug patterns and code design.
> 
> For this reason, I think a better approach is to put the unit tests
into a
> separate module, and rely on the internal access modifier to give them
> access to the "mostly private" data of the application code.
> 
> Gary
> 
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: IBM Linux Tutorials
> Free Linux tutorial presented by Daniel Robbins, President and CEO of
> GenToo technologies. Learn everything from fundamentals to system
> administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> _______________________________________________
> Nant-users mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/nant-users



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
Nant-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/nant-users

Reply via email to