ASDF with-muffled-compiler-conditions gripe`

Robert Goldman rpgoldman at sift.info
Mon Jun 11 01:25:48 UTC 2018


Well, if you are comfortable cutting that macro line out of your copy of 
ASDF, maybe you could tweak that line of code so that it prints the 
values of those variables?

I'm wondering if this is a bug in SBCL instead of ASDF.  I can replicate 
your result: on SBCL, tracing `COMPILE-FILE` does not print anything.  
*But* when I made this modification:
```
   (defun call-with-muffled-conditions (thunk conditions)
     "calls the THUNK in a context where the CONDITIONS are muffled"
     (handler-bind ((t #'(lambda (c) (when (match-any-condition-p c 
conditions)
                                       (format t "~&MUFFLING CONDITION 
~a~%" c)
                                       (muffle-warning c)))))
       (funcall thunk)))
```
... then I **did** see output from tracing `COMPILE-FILE`.

Also, on *neither* Allegro *nor* (my antique copy of) Clozure can I 
replicate this issue.

Pending new evidence, I think this is for the SBCL folks to deal with... 
  Is there any chance that SBCL might have trouble tracing calls inside 
`APPLY`?  I can't see why this macro would make any difference, now that 
I see that it's not trying to muffle anything.

Best,
r


On 9 Jun 2018, at 16:17, Cyrus Harmon wrote:

> Ah, right.
>
> 1. SBCL 1.4.8.53-eb16b4745 and ASDF 3.3.1
>
> 2. I'm not sure if these change when running asdf:load-system, but for 
> the moment they are all nil:
>
> CL-USER> asdf::*uninteresting-conditions*
> NIL
> CL-USER> asdf::*uninteresting-compiler-conditions*
> NIL
> CL-USER> asdf::*warnings-file-type*
> NIL
>
> thanks,
>
> Cyrus
>
> On 6/9/18 2:13 PM, Robert Goldman wrote:
>>
>> Two suggestions:
>>
>>  1. Let us know the implementation, implementation version, and ASDF
>>     version
>>  2. Look at the values of |*uninteresting-conditions*|,
>>     |*uninteresting-compiler-conditions*|, and |*warnings-file-type*|
>>     in your running lisp.
>>
>> That will help us figure out why you are seeing this effect.
>>
>> Thanks,
>> r
>>
>> On 9 Jun 2018, at 16:06, Cyrus Harmon wrote:
>>
>>     I don't know enough about ASDF to know if I'm running with
>>     deferred warnings -- I'm just trying to do (asdf:load-system ...)
>>     and removing the with-muffled-compiler-conditions form (which
>>     itself is inside a with-saved-deferred-warnings form) makes the
>>     trace output reappear. I could well be overestimating of that
>>     particular form, but somehow the TRACE output is being swallowed
>>     unless that form is removed.
>>
>>     On 6/9/18 1:54 PM, Robert Goldman wrote:
>>>
>>>     On 9 Jun 2018, at 12:38, Cyrus Harmon wrote:
>>>
>>>         Dear ASDF folks,
>>>
>>>         I was trying to trace calls to compile-file today and 
>>> noticed
>>>         that the with-muffled-compiler-conditions form in
>>>         uiop:compile-file* muffles trace output. Is there anyway to
>>>         make it not do that?
>>>
>>>         thanks,
>>>
>>>         Cyrus
>>>
>>>     I think you are overestimating the effect of this form. *UNLESS*
>>>     you are running with deferred warnings, this will only muffle
>>>     |*uninteresting-conditions*| and
>>>     |*uninteresting-compiler-conditions*|.
>>>
>>>     By default, those are bound to |NIL|, so this shouldn't be
>>>     causing your problem.
>>>
>>>     R
>>>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/asdf-devel/attachments/20180610/d33dafba/attachment.html>


More information about the asdf-devel mailing list