Re: RFC: A Test For Every Patch

I'd like to reject the philosophy. I'd prefer that the language define
perl. I.e. not an implementation. Rather the spec. 

Shouldn't the tests be designed from the spec? 

(Yes, it might depend if we can become machine independent (e.g. '%' or
NaN. Which is another forum.)

<chaim>

>>>>> "MGS" == Michael G Schwern <[email protected]> writes:

MGS> It has often be said that perl defines perl.  I'd like to change that a bit
MGS> and say perl's regression tests should define perl.  The state of the
MGS> existing tests is somewhat inadequate to the task.  Part of the problem is
MGS> tests are written whenever someone feels like adding a test.  

MGS> Instead, every new feature added/changed and bug fixed (ie. every code
MGS> patch) must have an associated patch to the testing suite.  No test, no
MGS> acceptance.  Period.

MGS> While this may seem draconian, it should rapidly raise the overall quality
MGS> of the patches by forcing authors to do more than dash off a little C code.


-- 
Chaim Frenkel					     Nonlinear Knowledge, Inc.
[email protected]				               +1-718-236-0183