Should we port or start from scratch?

Topics: Developer Forum
Coordinator
May 18, 2006 at 6:04 AM
One possibility for beginning coding is to start porting existing code from NUnit and remove functionality. The other is to start with a clean slate and create code, test-first I would expect.
Coordinator
May 18, 2006 at 6:06 AM
Continuing, cause I hit enter and it posted, which is annoying, and it won't let me edit a post, which is even more annoying. I'll post a bug on that.
Coordinator
May 18, 2006 at 6:09 AM
Sh*t! I did it again. OK, I'm in favor of starting from scratch and working test first. I think this will give us cleaner code, which we might eventually use to improve NUnit itself. Of course, we can always pull tests and code out of NUnit when it's convenient, but I think we should do it one item at a time rather than wholesale. If nothing else, this will make us use the source code management much more heavily so we can evaluate it.
Developer
May 18, 2006 at 7:49 AM
I think there's merit to the whole starting from 'scratch' idea...

It also gives us a streamlined way to re-evaluate even the most deeply intrenched stuff as a ramp-up to future NUnit developments...

The big question is what is the core purpose of NUnitLite?
Developer
May 18, 2006 at 1:43 PM
I guess I'm wondering the same thing. .NET already provides Debug.Assert which goes away whenever you compile to Release mode. Would developers still be able to run the NUnitGui against NUnitLite tests?
Developer
May 18, 2006 at 1:47 PM
Wow - the whole enter thing is annoying. ;) Continuing -- is the thought that people could ship NUnitLite with their code so if problems arose post-deployment they would have some way to run the user tests?
Coordinator
May 18, 2006 at 7:44 PM
Sorry, I deleted an aborted post that someone (Cory?) wrote and it said "Deleted 1 post and all it's responses" There were 8, now there are 6 posts. Another bug.
Developer
May 18, 2006 at 9:43 PM
I can also see merit in the starting from Scratch idea.

For the record, you can now press Enter and not have it automatically post, at least in IE.

I would suggest that we start from scratch except for our Interfaces, and here is why. I would like to see people be able to create Asserter extensions (implement IAsserter) and have that compilable for use on both NUnit and NUnitLite.

On the subject of what the purpose is, I see this for situations like what Charlie has cited previously where people have components which simply cannot work outside their pre-configured AppDomain.

I also feel strongly about it being compilable under the Compact Framework. This will open the door for someone to write a simple NUnitLite Gui on the mobile platforms. Such functionality works much better here when you plan ground-up to use the limited Compact Framework rather than in the full NUnit where the code was written for the full .NET Framework.
Coordinator
May 18, 2006 at 10:34 PM
Tim, it sounds as if you know a bit about the compact framework issues. Can you provide some background on the main diffs between CF and full .Net and between CF 1.0 and 2.0?