I've got two procedures that are nearly identical. They're only
different in the :data-argument in make-response. (I've simplified them
a bit to make them short and easy to read, but they're very close to the originals.)

--8<---------------cut here---------------start------------->8---
(defun head-response (r g i)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (article-headers (parse-article a))))))))

(defun body-response (r g i)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (article-body (parse-article a))))))))
--8<---------------cut here---------------end--------------->8---

Their differences are

head-response: (article-headers (parse-article a))
body-response: (article-body (parse-article a))

I wouldn't think I can express these as a /thunk/ because they depend on
the variable /a/ that is defined in a certain context in each procedure.

Is this the case for a macro? What do you advise? Thank you!

Keep in mind that I might write other procedures. For instance, here's
another one that's already written:

--8<---------------cut here---------------start------------->8---
(defun article-response (r g i)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (encode-body a))))))) --8<---------------cut here---------------end--------------->8---

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Julieta Shem <[email protected]> writes:

I've got two procedures that are nearly identical. They're only
different in the :data-argument in make-response.
...
Is this the case for a macro? What do you advise? Thank you!

A macro? Why? What's wrong with:

(defun foobar (r g i get-data)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (funcall get-data (parse-article a))))))))

- Alan

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2024-01-13, Julieta Shem <[email protected]> wrote:

I've got two procedures that are nearly identical. They're only
different in the :data-argument in make-response. (I've simplified them
a bit to make them short and easy to read, but they're very close to the originals.)

--8<---------------cut here---------------start------------->8---
(defun head-response (r g i)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (article-headers (parse-article a))))))))

(defun body-response (r g i)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (article-body (parse-article a))))))))
--8<---------------cut here---------------end--------------->8---

Here is one (untested) possibility:

(defmacro prepare-article-response (r g i avar expr)
`(let ((,avar (handler-case (fetch-article ,g ,i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep ,avar 'response) ,avar)
(t (prepend-response-with
(format nil "~a ~a" ,i (extract-mid ,avar))
(make-response :multi-line 'yes :code 200
:request ,r :data ,expr))))))

(defun head-response (r g i)
(prepare-article-response r g i a (article-headers (parse-article a))))

(defun body-response (r g i)
(prepare-article-response r g i a (article-body (parse-article a))))


The redundant calls to parse-article seem wasteful, but that's an organizational issue beyond the present scope.

Caveats: the i argument is evaluated twice, and r g i are not
evaluated in the order they appear in the macro argument list.

If this is only as a jig for writing functions like the above, where
those expressions are symbols referring to arguments, that's not a
problem.

We can write the macro by passing a lambda into a function:

(defun prepare-article-response-impl (r g i fun)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (funcall fun a)))))))

(defmacro prepare-article-response (r g i avar expr)
^(prepare-article-response-impl ,r ,g ,i (lambda (,avar) ,expr)))

defuns are as before. The order of evaluation issues disappear,
sine we are no inseting the argument expressions r g i into
a function argument list.


--
TXR Programming Language: http://nongnu.org/txr
Cygnal: Cygwin Native Application Library: http://kylheku.com/cygnal
Mastodon: @[email protected]
NOTE: If you use Google Groups, I don't see you, unless you're whitelisted.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 13 Jan 2024 19:46:09 -0300, Julieta Shem wrote:

Their differences are

head-response: (article-headers (parse-article a))
body-response: (article-body (parse-article a))

How about this (sorry, more of a Python programmer than a Lisp programmer,
but hopefully you get the idea):

(defun make-response (article-part)
(lambda (r g i)
(let
(
(a
(handler-case
(fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c))
)
)
)
(cond
((typep a 'response)
a
)
(t
(prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response
:multi-line 'yes
:code 200
:request r
:data (article-part (parse-article a))
)
)
)
) ; cond
) ; let
) ; lambda
) ; make-response

(fset head-response (make-response (symbol-function article-headers)))
(fset body-response (make-response (symbol-function article-part)))

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Alan Bawden <[email protected]> writes:

Julieta Shem <[email protected]> writes:

I've got two procedures that are nearly identical. They're only
different in the :data-argument in make-response.
...
Is this the case for a macro? What do you advise? Thank you!

A macro? Why? What's wrong with:

(defun foobar (r g i get-data)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (funcall get-data (parse-article a))))))))

That's pretty simple. Thanks. That's what I'm going to do here. I'll
leave the macro for when it's really needed --- learning to distinguish
when they're needed and when they're not.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Julieta Shem <[email protected]> writes:

That's pretty simple. Thanks. That's what I'm going to do here. I'll
leave the macro for when it's really needed --- learning to distinguish
when they're needed and when they're not.

The #2 trap that novice Lisp programmers fall into is using macros when
they don't need to. A novice that thinks they need to define a macro is probably wrong.

The #1 trap is calling EVAL. A novice that thinks they need to call
EVAL is always wrong.

- Alan

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Alan Bawden <[email protected]> writes:

Julieta Shem <[email protected]> writes:

That's pretty simple. Thanks. That's what I'm going to do here. I'll
leave the macro for when it's really needed --- learning to distinguish
when they're needed and when they're not.

The #2 trap that novice Lisp programmers fall into is using macros when
they don't need to. A novice that thinks they need to define a macro is probably wrong.

The #1 trap is calling EVAL. A novice that thinks they need to call
EVAL is always wrong.

Thanks for the advice. I haven't called eval yet, thankfully. :-) But I
think I can easily fall for #2. I'll post here all my macros to see
what you guys think.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Kaz Kylheku <[email protected]> writes:

On 2024-01-13, Julieta Shem <[email protected]> wrote:

I've got two procedures that are nearly identical. They're only
different in the :data-argument in make-response. (I've simplified them
a bit to make them short and easy to read, but they're very close to the
originals.)

--8<---------------cut here---------------start------------->8---
(defun head-response (r g i)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (article-headers (parse-article a))))))))

(defun body-response (r g i)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (article-body (parse-article a))))))))
--8<---------------cut here---------------end--------------->8---

Here is one (untested) possibility:

(defmacro prepare-article-response (r g i avar expr)
`(let ((,avar (handler-case (fetch-article ,g ,i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep ,avar 'response) ,avar)
(t (prepend-response-with
(format nil "~a ~a" ,i (extract-mid ,avar))
(make-response :multi-line 'yes :code 200
:request ,r :data ,expr))))))

Nice. I hadn't thought of the /avar/ idea.

(defun head-response (r g i)
(prepare-article-response r g i a (article-headers (parse-article a))))

(defun body-response (r g i)
(prepare-article-response r g i a (article-body (parse-article a))))

The redundant calls to parse-article seem wasteful, but that's an organizational issue beyond the present scope.

That's interesting. How would you avoid that redundancy? Any small
example would be nice.

We can write the macro by passing a lambda into a function:

(defun prepare-article-response-impl (r g i fun)
(let ((a (handler-case (fetch-article g i)
(sb-ext:file-does-not-exist (c)
(error-response c)))))
(cond ((typep a 'response) a)
(t (prepend-response-with
(format nil "~a ~a" i (extract-mid a))
(make-response :multi-line 'yes :code 200
:request r :data (funcall fun a)))))))

(defmacro prepare-article-response (r g i avar expr)
^(prepare-article-response-impl ,r ,g ,i (lambda (,avar) ,expr)))

That's pretty nice as well --- clears up a bunch of lambda expressions.

Thank you!

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 13 Jan 2024 23:08:05 -0000 (UTC), I wrote:

:data (article-part (parse-article a))

That “(article-part ...” should be “(funcall article-part ...”, since the
function is being passed as the value of a variable.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)