format called for effect

format called for effect

[Kai Großjohann]

When byte-compiling Gnus, I see the message from subject.  It comes
from gnus-msg.el, line 1497 or so, if I'm not mistaken.  I can't
reproduce the message by compiling the following simple file.  What's
the story?

(defun foo ()
  (condition-case ()
      (+ 3 4)
    (error (format "(setq foo %04d)\n" 42))))
-- 
Ambibibentists unite!

Re: format called for effect

[Lars Magne Ingebrigtsen]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> When byte-compiling Gnus, I see the message from subject.  It comes
> from gnus-msg.el, line 1497 or so, if I'm not mistaken. 

This code:

	  (condition-case ()
	      (pp `(setq ,(car olist)
			 ,(if (or (consp (setq sym (symbol-value (car olist))))
				  (and (symbolp sym)
				       (not (or (eq sym nil)
						(eq sym t)))))
			      (list 'quote (symbol-value (car olist)))
			    (symbol-value (car olist))))
		  (current-buffer))
	    (error
	     (format "(setq %s 'whatever)\n" (car olist))))

First of all -- I have no idea why that `condition-case' is the way
it is.  I didn't know that `format' had any effect?  Does it?  And if
it doesn't, why is that code that way?  If nobody has an explanation,
I'll just remove the `format'.
             
-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: format called for effect

[Kai Großjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:
>
>> When byte-compiling Gnus, I see the message from subject.  It comes
>> from gnus-msg.el, line 1497 or so, if I'm not mistaken. 
>
> This code:
>
> 	  (condition-case ()
> 	      (pp `(setq ,(car olist)
> 			 ,(if (or (consp (setq sym (symbol-value (car olist))))
> 				  (and (symbolp sym)
> 				       (not (or (eq sym nil)
> 						(eq sym t)))))
> 			      (list 'quote (symbol-value (car olist)))
> 			    (symbol-value (car olist))))
> 		  (current-buffer))
> 	    (error
> 	     (format "(setq %s 'whatever)\n" (car olist))))
>
> First of all -- I have no idea why that `condition-case' is the way
> it is.  I didn't know that `format' had any effect?  Does it?  And if
> it doesn't, why is that code that way?  If nobody has an explanation,
> I'll just remove the `format'.

Thinking some more about it, maybe what was intended was to try to
insert one setq statement, and to insert another setq statement in
the case of an error.

So maybe the condition-case should be around the if and return
`whatever' in case of error?
-- 
Ambibibentists unite!

Re: format called for effect

[Raja R Harinath]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:
>
>> When byte-compiling Gnus, I see the message from subject.  It comes
>> from gnus-msg.el, line 1497 or so, if I'm not mistaken. 
>
> This code:
>
> 	  (condition-case ()
> 	      (pp `(setq ,(car olist)
> 			 ,(if (or (consp (setq sym (symbol-value (car olist))))
> 				  (and (symbolp sym)
> 				       (not (or (eq sym nil)
> 						(eq sym t)))))
> 			      (list 'quote (symbol-value (car olist)))
> 			    (symbol-value (car olist))))
> 		  (current-buffer))
> 	    (error
> 	     (format "(setq %s 'whatever)\n" (car olist))))
>
> First of all -- I have no idea why that `condition-case' is the way
> it is.  I didn't know that `format' had any effect?  Does it?  

It doesn't.  And that seems to be the point of the overly cleverly
worded warning.  It think the warning says

  "you want a side-effect there, but 'format' ain't gonna do it for you"

- Hari
-- 
Raja R Harinath ------------------------------ harinath@cs.umn.edu

Re: format called for effect

[Lars Magne Ingebrigtsen]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> Thinking some more about it, maybe what was intended was to try to
> insert one setq statement, and to insert another setq statement in
> the case of an error.

Probably.  That's very, very old code, I think...

I've just removed the `error' bit and used `ignore-errors'.  It
doesn't make much sense to insert the `whenever' bits.

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: format called for effect

[Lars Magne Ingebrigtsen]

Raja R Harinath <harinath@cs.umn.edu> writes:

> It doesn't.  And that seems to be the point of the overly cleverly
> worded warning.  It think the warning says
>
>   "you want a side-effect there, but 'format' ain't gonna do it for you"

Yeah, probably.  It's a kinda strange thing to warn about -- did
anyone ever think that doing a `format' would have a side effect?

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: format called for effect

[Raja R Harinath]

Hi,

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Raja R Harinath <harinath@cs.umn.edu> writes:
>
>> It doesn't.  And that seems to be the point of the overly cleverly
>> worded warning.  It think the warning says
>>
>>   "you want a side-effect there, but 'format' ain't gonna do it for you"
>
> Yeah, probably.  It's a kinda strange thing to warn about -- did
> anyone ever think that doing a `format' would have a side effect?

It makes sense to warn in the case where you call to a pure (no
side-effects) function, and then discard the value.  Apparently, the
compiler figured out that the call to 'format' would be an expensive
no-op in that context, and that's why it complained.

It is just that the wording of the warning is wierd enough that you
put the blame on the callee (by worrying about 'format' in this case),
rather than the call site, where the problem actually is.

- Hari
-- 
Raja R Harinath ------------------------------ harinath@cs.umn.edu

Re: format called for effect

[Kai Großjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Yeah, probably.  It's a kinda strange thing to warn about -- did
> anyone ever think that doing a `format' would have a side effect?

Well...  But the byte-compiler *did* find a problem, didn't it?  So
the warning was actually a Good Thing, no?
-- 
Ambibibentists unite!

Re: format called for effect

[Lars Magne Ingebrigtsen]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> Well...  But the byte-compiler *did* find a problem, didn't it?  So
> the warning was actually a Good Thing, no?

Yup.  But, as noted, the error message could have been clearer.  Or,
perhaps, less clear, since what it said was 100% right, but perhaps a
bit too much to the point.  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen