[Mono-list] RFC: Corlib Unit Test on Linux How-To

[Nick Drochak]

How-To Run corlib Unit Tests on linux (1st draft)

These instructions are how I run the corlib unit tests on linux.  It's
not necessarily clean, nice, or pretty, but it works for me.  Comments
on this are welcome.  After a few iterations and edits, we'll add it to
the web site somewhere.

1) Get and build mono. See the www.go-mono.com site if you don't know
how to do this. I recommend the mono-build.sh build script available on
the download page.

2) Copy the class library dll's to the proper place.  "Proper place"
will depend on how you built mono, but it will need to be somewhere that
mono can find them. I put mine in the install/lib directory that the
script displays at the end. You'll need to build these on windows, or
get them from a pre-built package (see the web site for pointers). Here
is what you need to copy:
	mcs/class/lib/*.dll
	mcs/nunit/*.dll
	mcs/class/corlib/Test/corlib_res.dll

3) Copy mcs/nunit/NUnitConsole_mono.exe and
mcs/class/corlib/Test/corlib_linux_test.dll to the jit sub directory.
These are also built on Windows or from a pre-built package. Copy them
to mono/mono/jit/

4) Run Nunit in mono:
	cd mono/mono/jit
	./mono NUnitConsole_mono.exe
MonoTests.AllTests,corlib_linux_test.dll

If it's working you'll see some output like:

....E....F..F..E.EE..E.....E..F...F
.E.E...E..E.E..F...F..E...E.E.EE
(etc., etc)

Then most likely you'll get a stream of error and failure reports. The
difference is errors are unexpected problems with the tests and/or class
lib. Failures are tests that have failed, i.e. and Assert statement that
was not true. Both need to be fixed, but failures are easier to track
down.

Hopefully we can get more people trying these tests on linux, filing
bugzilla reports, or even better, posting patches to the list.

Comments, suggestions, and pizza are all welcome.

HTH,
Nick D.