bytecomp warning change

bytecomp warning change

[Richard Stallman]

I think this change is a bad idea.

+	* emacs-lisp/bytecomp.el (byte-compile-current-group): New var.
+	(byte-compile-file): Bind it.
+	(byte-compile-nogroup-warn): Use it to avoid spurious warnings when the
+	group argument is provided implicitly.
+	(byte-compile-format-warn, byte-compile-from-buffer)
+	(byte-compile-insert-header): Don't hardcode point-min==1.
+	(byte-compile-file-form-require): Remove unused var old-load-list.
+	(byte-compile-eval): Remove unused vars old-autoloads and hist-nil-new.

I want to have an explicit :group in every defcustom, so I put in a
warning when it is missing.  Did you just disable that warning?

Re: bytecomp warning change

[Stefan Monnier]

> +	* emacs-lisp/bytecomp.el (byte-compile-current-group): New var.
> +	(byte-compile-file): Bind it.
> +	(byte-compile-nogroup-warn): Use it to avoid spurious warnings when the
> +	group argument is provided implicitly.
> +	(byte-compile-format-warn, byte-compile-from-buffer)
> +	(byte-compile-insert-header): Don't hardcode point-min==1.
> +	(byte-compile-file-form-require): Remove unused var old-load-list.
> +	(byte-compile-eval): Remove unused vars old-autoloads and hist-nil-new.

> I want to have an explicit :group in every defcustom, so I put in a
> warning when it is missing.  Did you just disable that warning?

No, I did not disable it.
I only refined the warning to only complain when the group information is
really missing.


        Stefan

Re: bytecomp warning change

[Richard Stallman]

    No, I did not disable it.
    I only refined the warning to only complain when the group information is
    really missing.

The warning's purpose is to warn for implicitly specified groups, so
it is supposed to warn whenever the defcustom doesn't have a :group
keyword.  It looks like you've thrown out the baby and kept the
bathwater.

Would you please revert that part of the change?

Re: bytecomp warning change

[Stefan Monnier]

>     No, I did not disable it.  I only refined the warning to only complain
>     when the group information is really missing.

> The warning's purpose is to warn for implicitly specified groups, so
> it is supposed to warn whenever the defcustom doesn't have a :group
> keyword.  It looks like you've thrown out the baby and kept the
> bathwater.

Rather, what you consider as the baby is what I consider as the diapers.

> Would you please revert that part of the change?

Done,


        Stefan

Re: bytecomp warning change

[David Kastrup]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>     No, I did not disable it.  I only refined the warning to only complain
>>     when the group information is really missing.
>
>> The warning's purpose is to warn for implicitly specified groups, so
>> it is supposed to warn whenever the defcustom doesn't have a :group
>> keyword.  It looks like you've thrown out the baby and kept the
>> bathwater.
>
> Rather, what you consider as the baby is what I consider as the diapers.

Keeping the baby while throwing the diapers out may look like a good
idea at first.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: bytecomp warning change

[Stefan Monnier]

> Keeping the baby while throwing the diapers out may look like a good
> idea at first.

Believe me, you don't want to hold on to the diapers for too long.


        Stefan

Re: bytecomp warning change

[Richard Stallman]

    Rather, what you consider as the baby is what I consider as the diapers.

Letting :group default turned out to be unreliable, and I would
like to discourage the practice.

Re: bytecomp warning change

[Stefan Monnier]

>     Rather, what you consider as the baby is what I consider as the diapers.
> Letting :group default turned out to be unreliable,

I clearly disagree.

> and I would like to discourage the practice.

Yes, that much is clear.


        Stefan