Re: [elpa] externals/ebdb 4206ccd 02/21: Don't unilaterally add to kill-emacs-hook

Re: [elpa] externals/ebdb 4206ccd 02/21: Don't unilaterally add to kill-emacs-hook

[Stefan Monnier]

> +(defcustom ebdb-install-save-hook t
> +  "If non-nil, add `ebdb-save' to the hook `kill-emacs-hook'."
> +  :group 'ebdb-eieio
> +  :type 'boolean)

I think the docstring should talk about the intended effect from the
user's point of view, rather than from the implementation's point
of view.  I'm not 100% sure what is the intended effect, but it seems to
be something like "automatically save EBDB when leaving Emacs".


        Stefan

Re: [elpa] externals/ebdb 4206ccd 02/21: Don't unilaterally add to kill-emacs-hook

[Eric Abrahamsen]

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

>> +(defcustom ebdb-install-save-hook t
>> +  "If non-nil, add `ebdb-save' to the hook `kill-emacs-hook'."
>> +  :group 'ebdb-eieio
>> +  :type 'boolean)
>
> I think the docstring should talk about the intended effect from the
> user's point of view, rather than from the implementation's point
> of view.  I'm not 100% sure what is the intended effect, but it seems to
> be something like "automatically save EBDB when leaving Emacs".

Yes, that's the intended effect. I'd originally had the `add-hook' at
top level in ebdb-com.el, so loading the file added the hook. That led
to `ebdb-save' being called when batch compiling and testing (ugh). I'll
re-work the docstring.

Thanks,
Eric

Re: [elpa] externals/ebdb 4206ccd 02/21: Don't unilaterally add to kill-emacs-hook

[Eric Abrahamsen]

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

>> +(defcustom ebdb-install-save-hook t
>> +  "If non-nil, add `ebdb-save' to the hook `kill-emacs-hook'."
>> +  :group 'ebdb-eieio
>> +  :type 'boolean)
>
> I think the docstring should talk about the intended effect from the
> user's point of view, rather than from the implementation's point
> of view.  I'm not 100% sure what is the intended effect, but it seems to
> be something like "automatically save EBDB when leaving Emacs".

Actually, thinking in those terms, the option is poorly named, as well.

Re: [elpa] externals/ebdb 2b52686 18/21: Check type spec of slots that may or may not accept an object

[Stefan Monnier]

>     * ebdb.el (ebdb-field-bank-account): A type spec that looks like `(or
>       null class-name)' *must* specify the null first, then the class
>       name. Putting the class name first results in an error. This seems
>       like a bug to me -- it's an `or' statement, it shouldn't matter.

Maybe that deserves a bug report?


        Stefan

Re: [elpa] externals/ebdb 2b52686 18/21: Check type spec of slots that may or may not accept an object

[Eric Abrahamsen]

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

>>     * ebdb.el (ebdb-field-bank-account): A type spec that looks like `(or
>>       null class-name)' *must* specify the null first, then the class
>>       name. Putting the class name first results in an error. This seems
>>       like a bug to me -- it's an `or' statement, it shouldn't matter.
>
> Maybe that deserves a bug report?

I was going to find time to look at the code, and hopefully provide
both report and patch at the same time. I'm also investigating whether
it's possible to have nested customize interfaces in
`eieio-customize-object', when one object exists in the slot of another
-- it's almost certainly the same area of code.

I'll get to it!