help-gnu-emacs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Example use of macro to minimize and generalize the code


From: Jean Louis
Subject: Re: Example use of macro to minimize and generalize the code
Date: Tue, 1 Jun 2021 21:12:50 +0300
User-agent: Mutt/2.0.7+183 (3d24855) (2021-05-28)

* Yuri Khan <yuri.v.khan@gmail.com> [2021-06-01 20:58]:
> On Wed, 2 Jun 2021 at 00:27, Jean Louis <bugs@gnu.support> wrote:
> 
> > I have removed `id' argument and now I have:
> >
> > (macroexpand '(when-tabulated-id "people" (ignore)))
> > ⇒ (if id (progn (ignore))
> >     (if (or (eq "people" 'any) (string-equal "people" rcd-current-table))
> >         (let ((id (tabulated-list-get-id)))
> >           (if id (progn (ignore)) (message "Did not get ID")))
> >       (message "This function is for table `%s' only" "people")))
> >
> > If I would use ,d it would expand into number or nil, it seem it would end 
> > up same.
> >
> > Now I don't have compiler warnings, and functions work fine.
> >
> > Let me know if I am making errors in this.
> 
> Yes you are. Your macro expands to a form that refers to a symbol
> named “id” from outer scope. Next, there are several possibilities:
> 
> * The symbol is not defined, in which case the result of expansion
> will signal (void-variable id):
> 
>     (when-tabulated-id "people" (ignore))
>     ⇒ Debugger entered--Lisp error: (void-variable id)

Macro is exclusively used in functions as this:

(defun my-fun (&optional id)
  ""
  (interactive)
  (when-tabulated-id "hyobjects"
     ))

> * The user of your macro has to define a variable named “id” globally
> or buffer-locally. This is bad because the name “id” is not
> namespaced. In your own package, any variable names you define should
> start with a prefix unique to your package.

Do you mean all, or just global variables to have prefix?

>     (setq id t)
>     (when-tabulated-id "people" (ignore))
>     ⇒ nil

I understand, just macro is used exclusively in functions as above. Is
that alright?

> * The user has to use your macro in a context where “id” is bound as a
> local variable or a function argument. If that’s your intent, you
> should document it in the docstring of the macro, but it’s fragile
> because nobody expects forms to behave this way.
> 
>     (let ((id t))
>       (when-tabulated-id "people" (ignore)))
>     ⇒ nil
> 
>     (defun foo (id)
>       (when-tabulated-id "people" (ignore)))
>     (foo t)
>     ⇒ nil

(◍•ᴗ•◍) that can be, but that is exactly how I expected it to
behave. As the ID is either obtained programmatically which makes the
macro work or from (tabulated-list-get-id)

> * Lastly, you might have intended this macro to be used only by you.

Yes, and I see difficulties in future if somebody else wish to use it.

> In that case, it should be namespaced to your package with a double
> hyphen after the prefix. Anyway, three month later, you forget that
> the macro depends on “id” and you stare at your (let ((id t)) …) or
> (defun foo (id) …) which doesn’t obviously use “id” and remove the
> binding. Suddenly everything breaks.

Breaking I can also understand.

Macro or not macro my goal is to minimize programming, as 400+
functions are asking for (tabulated-list-get-id) and the ID can be in
different tables. 

Maybe it can be solved without any macro.

-- 
Jean

Take action in Free Software Foundation campaigns:
https://www.fsf.org/campaigns

Sign an open letter in support of Richard M. Stallman
https://stallmansupport.org/



reply via email to

[Prev in Thread] Current Thread [Next in Thread]