[elephant-devel] Re: elephant on ACL 6.2 using BDB

[Andrew Blumberg]

hi robert,

 	i wrote most of the MOP code originally, so i'll see if i can 
debug this.  i probably won't start looking at this until the weekend, but 
i'll write updates as soon as i get going.

 								- andrew

On Thu, 22 Dec 2005, Robert L. Read wrote:

> Deare Elephant Users and Developers,
>    Andrew Philpot has discovered a major bug in the 0.3.0 version of
> Elephant used on Allegro.
> As his (attached) debugging code shows, there is apparently a difference
> in the way Allegro and SBCL
> behave in terms of the MOP.
>
>    This means that at present, 0.3.0 doesn't work with Allegro and
> there is no simple workaround.
> I can't work on this much until after X-mas.  Fixing this is the highest
> priority for the project,
> in my opinion; I will work on it then.  However, I don't have a copy of
> Allegro and don't have
> much experience with the MOP.  If someone knows why Allegro "unsets" the
> slot values after
> they have been set and knows how to fix it, please tell me.
>
>    I will eventually solve this problem but it would be even better if
> a user could submit a patch
> for me.
>
>    Many thanks to Andrew for solving this problem; I will update the
> project home page
> to reflect this today.
>    As far as I know, 0.2.0 continues to work with Allegro.
>
>
>
> On Wed, 2005-12-21 at 17:44 -0800, Andrew Philpot wrote:
>
>> I agree that it seems to be some kind of ACL MOP difference.
>>
>> I metered various bits of initialization code I could find to see who
>> is setting slot DBCONNECTION-SPEC-PST of the persistent object.
>>
>> It looks like it gets set by all the specializations and then unset by
>> STANDARD-OBJECT's INITIALIZE-INSTANCE method.  I can't test that
>> directly, since I can't advice or specialize that CL-provided method.
>> If this is indeed what is happening, I think you will have an idea of
>> what to do, or perhaps we can formulate a question to Franz.
>>
>> Sorry for the messiness of the transcript, as I've run out of time.
>>
>> Thanks for taking a look.
>>
>> Andrew
>
>
>