gmane.lisp.ansi-tests

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T20:59:35

On Sun, Jan 8, 2012 at 9:55 PM, Ville Voutilainen
<ville.voutilainen< at >gmail.com> wrote:

Same here. Thanks for the explanation!

Bye,

Erik.

Re: Does this work at all?

Ville Voutilainen — 2012-01-08T20:55:57

On 8 January 2012 22:48, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll< at >googlemail.com> wrote:

Understood. Thanks for the explanation, and from my part, feel free to
go ahead and
modify the test as you indicated.

Re: Does this work at all?

Juan Jose Garcia-Ripoll — 2012-01-08T20:48:05

On Sun, Jan 8, 2012 at 9:33 PM, Ville Voutilainen <
ville.voutilainen< at >gmail.com> wrote:


It is not only that the assignment is executed in case of error: the
assignment to a constant is an incorrect form and does not need to be
accepted by the Common Lisp implementation. ECL rejects it beforehand and
thus cannot swallow that test.

Juanjo

Does this work at all?

Ville Voutilainen — 2012-01-08T20:33:07

Forgot to reply to the list...

On 8 January 2012 22:16, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll< at >googlemail.com> wrote:

Ah, now I see it: "If the store-value restart is invoked, its argument
becomes the new test-key, and is stored in keyplace as if by (setf
keyplace test-key). Then ctypecase starts over, considering each
clause anew."

The test didn't look like it's testing anything but the :good case,
but you are most certainly
correct that if the test for :keyword should fail, the test will do
funny things.


Seems reasonable.

Re: Does this work at all?

Juan Jose Garcia-Ripoll — 2012-01-08T20:16:55

First of all, Erik, thanks for granting access, but I would need some help
about how the repository is set up for others to contribute --
common-lisp.net does not provide info about it.

On Sun, Jan 8, 2012 at 8:33 PM, Erik Huelsmann <ehuels< at >gmail.com> wrote:



I would like to know on what basis you believe that the test is correct. It
is producing an incorrect form, which is an assignment to a constant. The
situation under those circumstances is unspecified The fact that ABCL had a
problem with the interaction between macroexpansion and the CASE statement
does not mean that the test is correct. Tests have to be kept if they
 conform to the ANSI specification and I am making a case that this test is
not. A proper form for that test would be

(PROGN
(DEFPARAMETER *CTYPECASE-15* :FOO)
(MACROLET ((%M (Z) Z))
 (CTYPECASE (EXPAND-IN-CURRENT-ENV (%M *CTYPECASE-15*))
   (INTEGER :BAD1)
   (KEYWORD :GOOD)
   (SYMBOL :BAD2)))
)

Juanjo

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T19:33:48

As you'll understand, the mail wasn't finished and prematurely sent.

On Sun, Jan 8, 2012 at 8:23 PM, Erik Huelsmann <ehuels< at >gmail.com> wrote:

of the ansi test suite.

Any implementer is welcome on this list, project and repository. Let
that be another invitation to anybody who's watching this thread and
has always felt left out.

Also, anybody who is Common Lisp Savvy (quite a bit) and wants to
extend the test suite is welcome.



As Ville pointed out, many CL savvy people are on that channel. But
among the CL savvy people there, there are also a number of
maintainers of other implementatiions. SBCL, CMUCL, ABCL and Allegro
come to mind. So, if there's no immediate reaction here, you
definitely could try in the IRC channel.

 Personally, I think the test really tests what it wants to and as
Ville points out, ABCL has had a problem in the past with one of the
(C)TYPECASE tests. I'm glad it was there. Given Ville's explanation of
what the test is testing (regardless that it does so with code that
can't be executed to completion succesfully), do you have a proposal
to replace it with?

I think it would be a pitty to start removing tests from the testsuite
without replacing those tests with acceptable variants: we'd run the
risk of eroding the test suite which for many implementers is a
valuable sanity check.


Regards,

Erik.

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T19:23:12

Hi Juan,



Done. I have offered access to most maintainers. Not all of them
wanted it or saw use centralize maintenance of 'their copy'


_______________________________________________
ansi-test-devel mailing list
ansi-test-devel< at >common-lisp.net
http://lists.common-lisp.net/cgi-bin/mailman/listinfo/ansi-test-devel

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T20:59:35

On Sun, Jan 8, 2012 at 9:55 PM, Ville Voutilainen
<ville.voutilainen< at >gmail.com> wrote:

Same here. Thanks for the explanation!

Bye,

Erik.

Re: Does this work at all?

Ville Voutilainen — 2012-01-08T20:55:57

On 8 January 2012 22:48, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll< at >googlemail.com> wrote:

Understood. Thanks for the explanation, and from my part, feel free to
go ahead and
modify the test as you indicated.

Re: Does this work at all?

Juan Jose Garcia-Ripoll — 2012-01-08T20:48:05

On Sun, Jan 8, 2012 at 9:33 PM, Ville Voutilainen <
ville.voutilainen< at >gmail.com> wrote:


It is not only that the assignment is executed in case of error: the
assignment to a constant is an incorrect form and does not need to be
accepted by the Common Lisp implementation. ECL rejects it beforehand and
thus cannot swallow that test.

Juanjo

Does this work at all?

Ville Voutilainen — 2012-01-08T20:33:07

Forgot to reply to the list...

On 8 January 2012 22:16, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll< at >googlemail.com> wrote:

Ah, now I see it: "If the store-value restart is invoked, its argument
becomes the new test-key, and is stored in keyplace as if by (setf
keyplace test-key). Then ctypecase starts over, considering each
clause anew."

The test didn't look like it's testing anything but the :good case,
but you are most certainly
correct that if the test for :keyword should fail, the test will do
funny things.


Seems reasonable.

Re: Does this work at all?

Juan Jose Garcia-Ripoll — 2012-01-08T20:16:55

First of all, Erik, thanks for granting access, but I would need some help
about how the repository is set up for others to contribute --
common-lisp.net does not provide info about it.

On Sun, Jan 8, 2012 at 8:33 PM, Erik Huelsmann <ehuels< at >gmail.com> wrote:



I would like to know on what basis you believe that the test is correct. It
is producing an incorrect form, which is an assignment to a constant. The
situation under those circumstances is unspecified The fact that ABCL had a
problem with the interaction between macroexpansion and the CASE statement
does not mean that the test is correct. Tests have to be kept if they
 conform to the ANSI specification and I am making a case that this test is
not. A proper form for that test would be

(PROGN
(DEFPARAMETER *CTYPECASE-15* :FOO)
(MACROLET ((%M (Z) Z))
 (CTYPECASE (EXPAND-IN-CURRENT-ENV (%M *CTYPECASE-15*))
   (INTEGER :BAD1)
   (KEYWORD :GOOD)
   (SYMBOL :BAD2)))
)

Juanjo

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T19:33:48

As you'll understand, the mail wasn't finished and prematurely sent.

On Sun, Jan 8, 2012 at 8:23 PM, Erik Huelsmann <ehuels< at >gmail.com> wrote:

of the ansi test suite.

Any implementer is welcome on this list, project and repository. Let
that be another invitation to anybody who's watching this thread and
has always felt left out.

Also, anybody who is Common Lisp Savvy (quite a bit) and wants to
extend the test suite is welcome.



As Ville pointed out, many CL savvy people are on that channel. But
among the CL savvy people there, there are also a number of
maintainers of other implementatiions. SBCL, CMUCL, ABCL and Allegro
come to mind. So, if there's no immediate reaction here, you
definitely could try in the IRC channel.

 Personally, I think the test really tests what it wants to and as
Ville points out, ABCL has had a problem in the past with one of the
(C)TYPECASE tests. I'm glad it was there. Given Ville's explanation of
what the test is testing (regardless that it does so with code that
can't be executed to completion succesfully), do you have a proposal
to replace it with?

I think it would be a pitty to start removing tests from the testsuite
without replacing those tests with acceptable variants: we'd run the
risk of eroding the test suite which for many implementers is a
valuable sanity check.


Regards,

Erik.

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T19:23:12

Hi Juan,



Done. I have offered access to most maintainers. Not all of them
wanted it or saw use centralize maintenance of 'their copy'


_______________________________________________
ansi-test-devel mailing list
ansi-test-devel< at >common-lisp.net
http://lists.common-lisp.net/cgi-bin/mailman/listinfo/ansi-test-devel

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T20:59:35

On Sun, Jan 8, 2012 at 9:55 PM, Ville Voutilainen
<ville.voutilainen< at >gmail.com> wrote:

Same here. Thanks for the explanation!

Bye,

Erik.

Re: Does this work at all?

Ville Voutilainen — 2012-01-08T20:55:57

On 8 January 2012 22:48, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll< at >googlemail.com> wrote:

Understood. Thanks for the explanation, and from my part, feel free to
go ahead and
modify the test as you indicated.

Re: Does this work at all?

Juan Jose Garcia-Ripoll — 2012-01-08T20:48:05

On Sun, Jan 8, 2012 at 9:33 PM, Ville Voutilainen <
ville.voutilainen< at >gmail.com> wrote:


It is not only that the assignment is executed in case of error: the
assignment to a constant is an incorrect form and does not need to be
accepted by the Common Lisp implementation. ECL rejects it beforehand and
thus cannot swallow that test.

Juanjo

Does this work at all?

Ville Voutilainen — 2012-01-08T20:33:07

Forgot to reply to the list...

On 8 January 2012 22:16, Juan Jose Garcia-Ripoll
<juanjose.garciaripoll< at >googlemail.com> wrote:

Ah, now I see it: "If the store-value restart is invoked, its argument
becomes the new test-key, and is stored in keyplace as if by (setf
keyplace test-key). Then ctypecase starts over, considering each
clause anew."

The test didn't look like it's testing anything but the :good case,
but you are most certainly
correct that if the test for :keyword should fail, the test will do
funny things.


Seems reasonable.

Re: Does this work at all?

Juan Jose Garcia-Ripoll — 2012-01-08T20:16:55

First of all, Erik, thanks for granting access, but I would need some help
about how the repository is set up for others to contribute --
common-lisp.net does not provide info about it.

On Sun, Jan 8, 2012 at 8:33 PM, Erik Huelsmann <ehuels< at >gmail.com> wrote:



I would like to know on what basis you believe that the test is correct. It
is producing an incorrect form, which is an assignment to a constant. The
situation under those circumstances is unspecified The fact that ABCL had a
problem with the interaction between macroexpansion and the CASE statement
does not mean that the test is correct. Tests have to be kept if they
 conform to the ANSI specification and I am making a case that this test is
not. A proper form for that test would be

(PROGN
(DEFPARAMETER *CTYPECASE-15* :FOO)
(MACROLET ((%M (Z) Z))
 (CTYPECASE (EXPAND-IN-CURRENT-ENV (%M *CTYPECASE-15*))
   (INTEGER :BAD1)
   (KEYWORD :GOOD)
   (SYMBOL :BAD2)))
)

Juanjo

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T19:33:48

As you'll understand, the mail wasn't finished and prematurely sent.

On Sun, Jan 8, 2012 at 8:23 PM, Erik Huelsmann <ehuels< at >gmail.com> wrote:

of the ansi test suite.

Any implementer is welcome on this list, project and repository. Let
that be another invitation to anybody who's watching this thread and
has always felt left out.

Also, anybody who is Common Lisp Savvy (quite a bit) and wants to
extend the test suite is welcome.



As Ville pointed out, many CL savvy people are on that channel. But
among the CL savvy people there, there are also a number of
maintainers of other implementatiions. SBCL, CMUCL, ABCL and Allegro
come to mind. So, if there's no immediate reaction here, you
definitely could try in the IRC channel.

 Personally, I think the test really tests what it wants to and as
Ville points out, ABCL has had a problem in the past with one of the
(C)TYPECASE tests. I'm glad it was there. Given Ville's explanation of
what the test is testing (regardless that it does so with code that
can't be executed to completion succesfully), do you have a proposal
to replace it with?

I think it would be a pitty to start removing tests from the testsuite
without replacing those tests with acceptable variants: we'd run the
risk of eroding the test suite which for many implementers is a
valuable sanity check.


Regards,

Erik.

Re: Does this work at all?

Erik Huelsmann — 2012-01-08T19:23:12

Hi Juan,



Done. I have offered access to most maintainers. Not all of them
wanted it or saw use centralize maintenance of 'their copy'


_______________________________________________
ansi-test-devel mailing list
ansi-test-devel< at >common-lisp.net
http://lists.common-lisp.net/cgi-bin/mailman/listinfo/ansi-test-devel