gmane.lisp.cmucl.general

Re: When slots :type are used

Raymond Toy — 2008-12-01T18:15:34

dvl> Hi again, dvl> Section "2.23.2 Slot Type Checking" continues to puzzle me, especially dvl> the fact that you have to be within *methods* for the :type slot option dvl> to be taken in consideration when writing something to a slot. For dvl> instance, with: dvl> (defclass foo () dvl> ((a :type fixnum))) dvl> (defvar *object* (make-instance 'foo)) dvl> (setf (slot-value *object* 'a) 3.5) ;; this works like a charm I think I have this fixed now: (defclass foo () ((a :type fixnum))) (defclass bar () ((a :type fixnum :initform 0))) (make-instance 'foo) -> signals error because a isn't initialized with a fixnum (defvar *object* (make-instance 'bar)) (setf (slot-value *object* 'a) 3.5) -> signals error I will need to run some tests and stuff, but if all goes well, it will be included soon (but probably not in time for the Dec snapshot). Ray

Re: When slots :type are used

Raymond Toy — 2008-12-01T15:50:44

Nicolas> Raymond Toy gmail.com> writes: >>> (defmethod setslot ((foo foo) val) >>> (setf (slot-value foo 'a) val)) >>> >>> (setslot *object* 3.5) ;; triggers a type error >>> >>> Can somebody provide a rationale for this ?? >>> >> Most likely because no got around to doing it. Plus the most >> interesting/most common case is probably in a method. Nicolas> Wasn't this introduced by Gerd Moellmann's PCL improvements? I remember Yes, Gerd made these changes. Ray

Re: When slots :type are used

Nicolas Neuss — 2008-11-27T08:42:00

Wasn't this introduced by Gerd Moellmann's PCL improvements? I remember that one of the characteristics of that change was that slot access inside a defmethod was treated specially and compiled to much faster code. (In an ideal world, this would have been an improvement of the compiler, but this was probably too hard to do.) Nicolas

Re: When slots :type are used

Raymond Toy — 2008-11-26T22:08:56

Most likely because no got around to doing it. Plus the most interesting/most common case is probably in a method. I'll see what I can do, but I know very little about the pcl implementation. Ray

When slots :type are used

Didier Verna — 2008-11-26T17:53:02

Hi again, Section "2.23.2 Slot Type Checking" continues to puzzle me, especially the fact that you have to be within *methods* for the :type slot option to be taken in consideration when writing something to a slot. For instance, with: (defclass foo () ((a :type fixnum))) (defvar *object* (make-instance 'foo)) (setf (slot-value *object* 'a) 3.5) ;; this works like a charm but: (defmethod setslot ((foo foo) val) (setf (slot-value foo 'a) val)) (setslot *object* 3.5) ;; triggers a type error Can somebody provide a rationale for this ?? Thanks.

:type slot option in classes

Didier Verna — 2008-11-26T09:39:47

Hello, Reading section "2.23.2 Slot Type Checking" of the manual, I'm inclined to think that the following code should trigger an error when instanciating the class. But it doesn't: | (declaim (optimize (speed 0) | (compilation-speed 0) | (safety 3) | (debug 0))) | | | (defclass foo () | ((slot :type single-float :initform "bar"))) The behavior is not undefined (as the standard would say), but coherent in the sense that objects of any type can be correctly written and read from the slot, just as if the :type option was not given. Am I missing something ? Thanks! BTW, some more info: * pcl::*use-slot-types-p* T * (lisp-implementation-version) "CVS 19d 19d-release (19D)"

Re: compilation note -> style-warning?

Raymond Toy — 2008-11-25T15:03:26

I think changing the efficiency notes (and the boxing notes in particular) to be an efficiency-warning would not be hard. Making efficiency-warning not be a subtype of style-warning is easy too. But I have not given any thought about the implications of such a hierarchy. Ray

Re: compilation note -> style-warning?

Nicolas Neuss — 2008-11-25T14:52:31

This issue shows up in my Femlisp test suite when runtime-compiling some scalar-product operation. My test function looks like this at the moment (defun test-function (func) (catch 'trap (handler-bind ((serious-condition #'(lambda (condition) (throw 'trap condition))) #-(or cmu scl) (warning #'(lambda (condition) (throw 'trap condition)))) (funcall func) nil))) So you see that I can live with the current behaviour, but it would be nice if I could drop the conditional for CMUCL. For me, the best thing would be if efficiency notes would not be style warnings, although I understand that this might be a too large change for a minor improvement. (Is that subtyping at least useful in some other situation?) Regards, Nicolas

Re: compilation note -> style-warning?

Raymond Toy — 2008-11-25T14:29:08

Nicolas> Hello, Nicolas> in the following code a compilation note (returning of double) triggers Nicolas> also a style-warning. Is this intended? (I cannot see how to improve the Nicolas> style...) Not sure. Maybe compilation notes like this should not be a style-warning. It's really an efficiency note, but currently efficiency notes are also style-warnings. Would an efficiency-warning, which would be a subtype of style-warning work for you? (I think there might also a small bug. The note is a compiler-note, but the routine that generates it is called do-coerce-efficiency-note.) Ray

compilation note -> style-warning?

Nicolas Neuss — 2008-11-25T10:11:35

Hello, in the following code a compilation note (returning of double) triggers also a style-warning. Is this intended? (I cannot see how to improve the style...) (catch 'trap (handler-bind ((serious-condition #'(lambda (condition) (throw 'trap condition))) (warning #'(lambda (condition) (throw 'trap condition)))) (compile nil (lambda (x n) (declare (type fixnum n) (type (simple-array double-float (*)) x)) (declare (optimize (safety 0) (space 0) (debug 0) (speed 3))) (loop for i of-type fixnum from 0 below n summing (* (aref x i) (aref x i)) of-type double-float))))) Yours, Nicolas

Re: Please explain this compiler note

Stelian Ionescu — 2008-11-19T17:57:59

the latter

Re: Please explain this compiler note

Didier Verna — 2008-11-19T17:45:34

Hmmm, I'm confused. The return value of DEFUN is the function name, right? Or do you mean that when CMU-CL creates the accessor function, it generates code to box the return value and /that/ is what triggers the note?

Re: Please explain this compiler note

Stelian Ionescu — 2008-11-19T16:58:22

On Wed, 2008-11-19 at 16:40 +0100, Didier Verna wrote: The return value of (DEFUN MYSTRUCT-SLOT (OBJECT) ...) : the value of the slot SLOT is boxed hence the compiler note.

Please explain this compiler note

Didier Verna — 2008-11-19T15:40:18

[Fwd: ILC]

Raymond Toy — 2008-10-29T13:04:44

In case you didn't already know, there's another Lisp conference coming up. Ray, who might go if he plans things out better this time than he did for OOPSLA.

Re: Mailing list issues?

Raymond Toy — 2008-10-16T18:54:52

Could you try that? The issue is the newly subscribed addresses do not receive messages from the list and cannot send messages to the list. (I assume the messages are sent, but they never appear on the list.) Thanks, Ray

Re: Mailing list issues?

Raymond Toy — 2008-10-16T18:41:32

Interesting. I have not (yet?) received any message from you from your gmail account. Ray

Re: Mailing list issues?

Martin Cracauer — 2008-10-16T17:59:54

Dunno, folks, works fine for me. I just subscribed to cmucl-imp using my gmail account. Subscription went smooth and I have received the test message I sent a couple minutes ago at that newly subscribed address. If you send me a list of problematic addresses I can slam them into listar by hand. Martin

Re: Mailing list issues?

Raymond Toy — 2008-10-16T17:36:47

Yes, that's my understanding. People subscribe, and they get mails saying they're subscribed. But they don't receive messages and can't send messages. Also, I had sent (private) email earlier to you at your cons.org address. Is that no longer really used? Is your itasoftware one better? Thanks for looking into this. Ray

Re: Mailing list issues?

Martin Cracauer — 2008-10-16T15:12:31

I'll look into this. All these spamfilter and security updates wreck the lists every now and then. I receive messesages just fine. Do you say that subscription is broken? Martin Raymond Toy wrote on Thu, Oct 16, 2008 at 10:50:36AM -0400:

Re: Mailing list issues?

Raymond Toy — 2008-10-13T15:42:23

Just a clarification. The mailing list appears ok, if you've previously received messages. But new subscribers appear to have problems. listar (or the mailing list software) says they're subscribed, but new subscribers apparently cannot send or receive messages, despite being subscribed. Ray