Re: Oort Gnus v0.19 is released

[Kevin Greiner <kgreiner@xpediantsolutions.com>]

Jesper Harder <harder@myrealbox.com> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> Jesper Harder <harder@myrealbox.com> writes:
>>
>>> There is a grain of truth in the warnings, i.e. that pop is a tiny bit
>>> less efficient when we don't use the return value -- not that I think
>>> anyone would notice the difference.
>>
>> The warning has been removed, but I'm not sure I agree it had even a
>> tiny grain of truth in it -- the docstring for `pop' says "Remove and
>> return the head of the list stored in PLACE.", and I assume this is
>> what CL `pop' is supposed to do.  Using it in that way in elisp should
>> never generate a warning IMHO.
>
> I agree that the warning is too pedantic.  But it's still true,
> though.  The part to notice in the docstring is
>
>    and return the head of the list
>  
> If we don't use the return value from pop, then
>
>   (setq x (cdr x))
>
> is obviously more efficient than 
>
>   (pop x) == (car (prog1 x (setq x (cdr x))))

Of course, the optimizer should (after printing the warning) remove
the car and prog1 leaving just (setq x (cdr x)).

That being said, I'll probably remove the pop macros in gnus-agent.

Kevin