[Pharo-project] Issue 3270 in pharo: fix xor: to work with blocks and booleans
Status: Fixed Owner: marcus.denker Labels: Milestone-1.2 New issue 3270 by marcus.denker: fix xor: to work with blocks and booleans http://code.google.com/p/pharo/issues/detail?id=3270 ... so that it works like or: (calling #value on the argument. This means it will continue to work with booleans, too). Attachments: XOr.1.cs 903 bytes
Hi, I don't get why it should work with blocks, since evaluation of the exclusive OR operation cannot be short-circuited. Therefore there is no reason to ever use a block as an argument. #and: and #or: take a block as an argument precisely because for 50% of the receivers (i.e., false and true, respectively) it is known in advance that the value of the expression in the argument block is irrelevant, so it would be at least wasteful to evaluate it. But this reasoning does not apply to #xor:. I think it is simply a user mistake when a block is provided as argument, where a boolean is required. It's an understandable mistake (that I've made myself ;) since xor: has such similarities to or: that you automatically want to start a block unless you're really thinking about what you're doing. So, I don't see the bug. Thus, I don't see the fix. I do see obscure code (aBoolean value)and wasted time because of the extra message send. Apologies if this is a late response, I unly just saw the issue mentioned. Cheers, Peter On Thu, 18 Nov 2010 18:55:41 +0100, <pharo@googlecode.com> wrote:
Status: Fixed Owner: marcus.denker Labels: Milestone-1.2
New issue 3270 by marcus.denker: fix xor: to work with blocks and booleans http://code.google.com/p/pharo/issues/detail?id=3270
... so that it works like or: (calling #value on the argument. This means it will continue to work with booleans, too).
Attachments: XOr.1.cs 903 bytes
+1 -- Cheers, Peter On Mon, Nov 29, 2010 at 10:49 AM, Peter van Rooijen <peter@vanrooijen.com>wrote:
Hi,
I don't get why it should work with blocks, since evaluation of the exclusive OR operation cannot be short-circuited. Therefore there is no reason to ever use a block as an argument.
#and: and #or: take a block as an argument precisely because for 50% of the receivers (i.e., false and true, respectively) it is known in advance that the value of the expression in the argument block is irrelevant, so it would be at least wasteful to evaluate it.
But this reasoning does not apply to #xor:.
I think it is simply a user mistake when a block is provided as argument, where a boolean is required.
It's an understandable mistake (that I've made myself ;) since xor: has such similarities to or: that you automatically want to start a block unless you're really thinking about what you're doing.
So, I don't see the bug. Thus, I don't see the fix. I do see obscure code (aBoolean value)and wasted time because of the extra message send.
Apologies if this is a late response, I unly just saw the issue mentioned.
Cheers, Peter
On Thu, 18 Nov 2010 18:55:41 +0100, <pharo@googlecode.com> wrote:
Status: Fixed
Owner: marcus.denker Labels: Milestone-1.2
New issue 3270 by marcus.denker: fix xor: to work with blocks and booleans http://code.google.com/p/pharo/issues/detail?id=3270
... so that it works like or: (calling #value on the argument. This means it will continue to work with booleans, too).
Attachments: XOr.1.cs 903 bytes
participants (3)
-
Peter Hugosson-Miller -
Peter van Rooijen -
pharo＠googlecode.com