On Fri, May 30, 2008 at 6:17 PM, Adrian Lienhard <adi@netstyle.ch> wrote:
No I don't agree. Because it is not likely that there is a test for Matrix that will show a potential bug. I just wanted to point out that it is important to precisely look at the effect of such "global" transformations. (E.g. like Edgar tried to replace all #== with #= IIRC ;)
With your proposed change, I think it is unlikely you will break anything, but you cannot say for sure without detailled *manual* checking. The difficulty is that you theoretically needed to verify that there is never passed a Matrix instance to "ReadStream on:".
I don't understand why you are choosing this example of Matrix. Since Matrix is not an Array nor a String, it is impossible that a code like 'ReadStream on: aMatrix' ever worked. In fact, this reasoning could be applied to any other class which is neither Array nor String or one of their subclasses.
Also to consider with such changes is that you may break external code that you don't know of.
How? If don't remove any method. Existing code should still work. Do you have any example in mind?
To sum up, I'm not against such improvements at all, but they have to be done carefully, i.e., the argument that tests will show any problem does not hold.
I knew the tests argument did not hold because there are so few tests in Squeak, I was joking :-). I'm trying to do the changes carrefully, that's why I commited things package after package instead of doing it globally. -- Damien Cassou Peter von der Ahé: «I'm beginning to see why Gilad wished us good luck». (http://blogs.sun.com/ahe/entry/override_snafu)