Re: [Sbcl-devel] Vicious error message

[Luke J C. <lu...@ba...>]

On Wed, 07 Oct 2009 11:57:41 -0700, Christophe Rhodes <cs...@ca...>  
wrote:

> "Luke J Crook" <lu...@ba...> writes:
>
>> On Wed, 07 Oct 2009 00:49:48 -0700, Christophe Rhodes
>> <cs...@ca...> wrote:
>>
>>> Luke J Crook <lu...@ba...> writes:
>>>
>>>> What does the following mean exactly? :)
>>>
>>> It means that something has caused the effective method (the  
>>> combination
>>> of all the applicable methods) for the generic function behind
>>>    (slot-value <x> 'lispbuilder-sdl::pause)
>>> to be recomputed: and that that computation needs to be able to call  
>>> the
>>> same generic function (which isn't possible, because we don't know to
>>> call it yet).  The system knows how to break many of these vicious
>>> metacircles, but not this one.
>>>
>>>> vicious metacircle:  The computation of an effective method of
>>>> #<STANDARD-GENERIC-FUNCTION (SB-PCL::SLOT-ACCESSOR :GLOBAL
>>>>                              LISPBUILDER-SDL::PAUSE
>>>>                              SB-PCL::READER) (1)>
>>>> for arguments of types (#<STANDARD-CLASS LISPBUILDER-SDL:MIXER>)
>>>> uses the effective method being computed.
>>>>    [Condition of type SIMPLE-ERROR]
>>>
>>> Having said that: I am surprised that this can even happen in this  
>>> case;
>>> you've got standard classes and standard generic functions everywhere.
>>> What weirdness is going on?  The same method called from multiple
>>> threads on startup?  Interposing MOP stuff?  I can't tell just from  
>>> this
>>> information.
>>
>> That gives me at least something to go on, thanks. I'm definitely not
>> trying to do anything fancy with the MOP, just the the most basic CLOS
>> possible to get the job done. The code in question is here if you are
>> interested. I'll spend some time and try to work this out.
>
> That does all look rather normal.  Is it repeatable?  How do you trigger
> it?  Can you get a backtrace?

There was no backtrace available (slime's own words). And each time I  
tried to exit to the REPL it threw me back into the same error until it  
eventually complained that it exceeded the max number of nested restarts  
(or something, sorry I forgot the actually message returned), and then  
dropped into ldb.

It hasn't happened since though.

- Luke