[Openmcl-devel] Weird ASDF problem

[Ron Garret]

On Sep 11, 2010, at 6:48 PM, Robert Goldman wrote:

> On 9/11/10 Sep 11 -1:53 PM, Ron Garret wrote:
>> 
>> On Sep 11, 2010, at 11:42 AM, Ron Garret wrote:
>> 
>>> I would suggest replacing (truename #p"ccl:") in wrapping-output-translations with (let ((*default-pathname-defaults* nil)) (truename "ccl:"))
>> 
>> That should be:
>> 
>> (let ((*default-pathname-defaults* #P"")) ...
>> 
>> not NIL.
>> 
>> I found that just putting this in my ccl-init.lisp seems to be an adequate workaround:
>> 
>> (let ((*default-pathname-defaults* #P"")) (asdf::initialize-output-translations))
>> 
>> rg
>> 
> 
> Is there any reason we shouldn't replace
> 
> #+clozure (,(wilden (truename #p"ccl:")) ())
> 
> with
> 
> #+clozure (,(wilden (let ((*default-pathname-defaults* #P""))
>                      (truename #p"ccl:"))) ())
> 
> in asdf.lisp?
> 

There might be someone out there who has a setting for *default-pathname-defaults* that plays nicely with (truename #p"ccl:") and whose behavior they are relying on.  This patch would break that.  Of course, there may not be any such people.  The community is not that big.

It would help if I understood what wrapping-output-translations was supposed to do.  There's a comment on that line that says:

 ; not needed: no precompiled ASDF system

so maybe the whole thing is, as the comment says, not needed :-)

> If you think that's right, we could make that change.

I'm really not the right person to make that call.

> It really would be best if you were to make a launchpad bug for this,
> though, so that other developers would see this.  I'd be more confident
> offering this as a patch if other ccl-using asdfers could see....

I've never used launchpad, and at first glance it doesn't seem to be very user friendly.  I would expect to see a "Report a bug" link or something like that.  If it's there, it's well hidden.

rg