[mcclim-devel] mcclim failure with asdf3

[Faré]

Works better with the patch attached.

Once again, this only fixes the build on clisp and allegro,
and solves a fiasco with *features*, but it doesn't address
the many warnings in the build, that some developer should address.

—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org
To have a right to do a thing is not at all the same
as to be right in doing it.     — G.K. Chesterton


On Wed, Feb 27, 2013 at 3:22 AM, Faré <fahree at gmail.com> wrote:
> Here is a patch to mcclim.asd that solves several issues with it.
> It does *not* solve the issue with warnings;
> someone who better understands McCLIM needs to address them.
>
> What it does is:
> * simplify the (require :clx) code for allegro and clisp.
>  The previous traverse methods caused an error on asdf3
>  because of a signature change, but were broken since asdf2 (years ago),
>  because of a semantic change in traverse (necessary to make its
>  complexity linear rather than cubic in the size of a system).
>  That was a showstopper. Instead, I do things in much simpler ways,
>  that work and are supported.
>
> * remove the braindead perform :around binding *features*,
>  that totally defeated the other perform :after methods that modified it.
>  Instead, don't bind *features*, but just push :mcclim on the features
>  earlier, early enough to compile ESA and Drei with it
>  (and yes, grepping through the code shows it works; the :building-mcclim
>  feature is always redundant with :mcclim; it should just die.)
>
> * all these uses of make-pathname were necessary in ASDF1,
>  but haven't been since ASDF2. I simplified them away.
>
> One of you still has to resolve all the warning issues.
> Some of them *might* disappear if you merge some of your systems together.
> Or not.
>
> —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org
> La vie n'est pas un restaurant mais un buffet. Levez-vous pour vous servir.
>         Life is not a restaurant but a buffet. Stand up to serve yourself.
>                 — Dominique Glocheux
>
>
> On Tue, Feb 26, 2013 at 9:18 PM, Faré <fahree at gmail.com> wrote:
>> Note that these methods on traverse that break the build on asdf3
>> had been totally ineffective since before asdf 2.000,
>> for my changes to make traverse performance linear rather than cubic
>> in the size of the system were such that these methods would likely
>> never be called.
>>
>> I can provide a working replacement, but I'm not going to bother unless
>> there's an active committer to commit my patch.
>>
>> —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org
>> As far as we know, our computer has never had an undetected error.
>>                 — Weisert
>>
>>
>> On Tue, Feb 26, 2013 at 8:54 PM, Anton Vodonosov <avodonosov at yandex.ru> wrote:
>>> In addition, mcclim and all the dependent systems fail on CLISP and Allegro with ASDF3
>>> because .asd file tries to define a method for asdf:traverse generic function.
>>>
>>> The method signature is incompatible with the generic function anymore -
>>> doesn't have &REST or &KEY.
>>>
>>> Best regards,
>>> - Anton
>>>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: asdf2.diff
Type: application/octet-stream
Size: 13991 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/mcclim-devel/attachments/20130227/f44cbe5a/attachment.obj>