Search
Wednesday, 12th of August 2020, 0:52:10 UTC
3:58:05
beach
Good morning everyone!
3:59:01
no-defun-allowed
Good morning beach!
4:10:09
beach
heisig: What does this mean: There is no applicable method for the generic function #<... TRUCLER-RESTRICT-FOR-MACROLET-EXPANDER (1)> when called with arguments (<client> <global environment>)? Should I define a method on that function for my global environment?
4:26:17
beach
It seems if I do (defmethod trucler:restrict-for-macrolet-expander ((client client) (environment environment)) environment) [in SICL-BOOT] then the message disappears and things seem to work.
5:14:34
no-defun-allowed
Good morning Harag
5:58:44
beach
Colleen: Weather in Bordeaux
5:58:46
Colleen
Weather in Bordeaux: Possible Light Rain at 20°C (feels like 21°C), 88% humidity, 2km/h wind, 1014hPa pressure.
5:59:05
beach
And thunder storms all day.
6:47:47
heisig
beach: I have never touched the generic function TRUCLER-RESTRICT-FOR-MARCOLET-EXPANDER.
6:48:34
heisig
I don't think so. Let me check...
6:49:09
beach
It is trucler:restrict... not trucler-restrict... obviously.
6:49:42
beach
Oh, it could be Bike.
6:50:48
heisig
No, you wrote it in commit 4e8004791d31e57d9990a1f27436b01b83850a41.
6:52:10
heisig
Placing an identity function there is probably good for now. But the method will have to do some more work to be fully correct.
6:52:50
heisig
I think this is about stripping out things like surrounding flet bindings that shouldn't be visible inside of macros bound via macrolet.
6:53:07
beach
I see, yes. I still think this is an idea by Bike. I must have just changed the name.
6:54:16
heisig
Oh, by the way, I just had an idea about representing the sign of a SICL bignum.
6:54:52
heisig
Why don't we introduce and use the classes NON-NEGATIVE-BIGNUM and NEGATIVE-BIGNUM?
6:55:32
beach
What are the pros and cons?
6:57:28
heisig
The pros are that it simplifies the implementation. The sign can be handled by specialized methods for the four cases of each binary function.
6:57:46
no-defun-allowed
Good morning heisig!
6:57:59
heisig
And it requires zero extra space per bignum. There is just a different value in the class slot of the header.
6:58:42
beach
That's true. Though the initial plan was to stick a bit in with the limb count.
6:58:51
beach
But that's definitely messy.
6:59:04
heisig
No, that is also a fine choice.
6:59:52
heisig
I just wanted to point out that we have the infrastructure to encode the sign directly in the class.
7:00:00
heisig
Good morning no-defun-allowed!
7:00:40
beach
heisig: Yes, and I think it is a good idea. I'll definitely give it some thought.
7:02:54
beach
I think Bike invented this restriction, but called it COMPILE-TIME. I must have found the name insufficiently informative.
Wednesday, 12th of August 2020, 12:52:10 UTC