[Mono-devel-list] Re: Test policy proposition

Atsushi Eno atsushi at ximian.com
Mon May 30 10:11:15 EDT 2005


I don't think your proposal changes anything. That's what we have
actually had.

Note that it not *me* to *accept* your proposal, since your
discussion is for the whole tests (and I don't think there is
any reason to limit the scope of discussion only for xml tests).

> I want to propose the idea of policy regarding running tests before 
> commit for preventing regressions. I have to go right now for a few 
> hours, so I simply post the draft of README, if it's not clear enough, 
> we'll discuss it when I return. If everything is fine for you, let me 
> know, and I'll commit the testsuit enforcing this policy for W3C tests.

If you believe that you have kept the rule and will keep it
for yourself, then feel free to add the "rule" to README. In that
case then there are some additional points:

	- it is required to run tests on mono runtime. I have
	  seen some patches that does not work fine under Mono

	- that rule applies to patches being posted. I have
	  seen some patches that has regressions.

	- We need some clarification on the "commit" you say here.
	  It is not obvious. Most of us will object to your idea
	  to enforce the tests before committing System.Double
	  or System.CodeDom.* changes that affects on XSLT.

> Notice, that this testsuite ignores regressions for tests that fail in 
> dot net. Wheither it must be so - I don't know, let's decide together.

Why did you make it so? We don't want to avoid those tests.

Though as for XSLT tests it still makes sense since our test
depends on MS output.

Atsushi Eno

More information about the Mono-devel-list mailing list