diff mbox series

[bug#67786] doc: Add documentation for define-record-type*

Message ID 9bf2efa9c7aab1661fcf5180d1e536fc6dc0e9b3.1702324538.git.skyvine@protonmail.com
State New
Headers show
Series [bug#67786] doc: Add documentation for define-record-type* | expand

Commit Message

Skyler Ferris Dec. 11, 2023, 8:16 p.m. UTC
* doc/guix.texi: Add sections describing the typical usage and API
reference for define-record-type*

Change-Id: I19e7220553d10652c794e6e0172b2c9ee961f54f
---
 doc/contributing.texi |   1 -
 doc/guix.texi         | 274 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 274 insertions(+), 1 deletion(-)


base-commit: 2b782f67266b42bb40015bd23ce2443be2f9b01f

Comments

Skyler Ferris Dec. 11, 2023, 8:19 p.m. UTC | #1
This is a duplicate of bug 67786. Apologies for the noise.

On 12/11/23 12:18, help-debbugs@gnu.org wrote:
> Thank you for filing a new bug report with debbugs.gnu.org.
>
> This is an automatically generated reply to let you know your message
> has been received.
>
> Your message is being forwarded to the package maintainers and other
> interested parties for their attention; they will reply in due course.
>
> Your message has been sent to the package maintainer(s):
>   guix-patches@gnu.org
>
> If you wish to submit further information on this problem, please
> send it to 67787@debbugs.gnu.org.
>
> Please do not send mail to help-debbugs@gnu.org unless you wish
> to report a problem with the Bug-tracking system.
>
> --
> 67787: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=67787
> GNU Bug Tracking System
> Contact help-debbugs@gnu.org with problems
Skyler Ferris Dec. 28, 2023, 2:03 a.m. UTC | #2
It looks like the non-breaking space issue is still not resolved, and 
I'm not immediately seeing how to fix it. I'll look at it more later. My 
`git send-email` usage has been failing to deliver even though it worked 
for previous patches, so I tried pasting the email into Thuderbird to 
send it manually, but it seems like Thunderbird is stubborn about about 
modifying the contents of my email automatically. It will probably work 
best to fix the problem with `git send-email`.
Skyler Ferris Dec. 29, 2023, 12:12 a.m. UTC | #3
Hello again,

I'm not sure if the problem is on my end. I reviewed the logs for the 
initial message I sent on December 11 and for the one I tried to send 
yesterday. In both cases, exim indicates that it connected to the 
appropriate endpoint, wrote a chunked message (BDAT), concluded with a 
line like "SMTP<< 250 2.0.0 Ok: <number> bytes queued as <id>", then 
closed the connection. The only difference is that when sending to 
guix-patches@gnu.org exim ended up talking to eggs.gnu.org, but when 
sending to this bug's email address it talked to debbugs.gnu.org. Is 
there someone I can talk to in order to help diagnose this issue?

Thanks,
Skyler
Skyler Ferris Dec. 29, 2023, 12:17 a.m. UTC | #4
Oh, I see that there is a help-debbugs mailing list. I'll reach out there.
diff mbox series

Patch

diff --git a/doc/contributing.texi b/doc/contributing.texi
index 9e9b89782c..60fcf95b77 100644
--- a/doc/contributing.texi
+++ b/doc/contributing.texi
@@ -1311,7 +1311,6 @@  Data Types and Pattern Matching
 notably the fact that it is hard to read, error-prone, and a hindrance
 to proper type error reports.
 
-@findex define-record-type*
 @findex match-record
 @cindex pattern matching
 Guix code should define appropriate data types (for instance, using
diff --git a/doc/guix.texi b/doc/guix.texi
index 1fd2e21608..e9d0fd1466 100644
--- a/doc/guix.texi
+++ b/doc/guix.texi
@@ -12561,6 +12561,280 @@  G-Expressions
 @code{(*approximate*)}, but this may change.
 @end deffn
 
+@node Records in Guix
+@section Records in Guix
+Guix uses @code{define-record-type*} to define structures with a lispy format.
+Packages, operating systems, etc are all defined with
+@code{define-record-type*} facilities. If one was using this facility to
+define preferences for a text editor, it might look like this:
+
+@lisp
+;; The only valid emulation modes are the symbol 'emacs, the symbol 'vim, or
+;; the boolean #f. As a convenience to the user, if they pass in a string
+;; first convert it to a symbol and accept it if it is valid.
+(define (sanitize-emulation-mode value)
+  (let ((symbolized-value (cond ((not value)     #f)
+                                ((string? value) (string->symbol value))
+                                (#t              value))))
+    (unless (or (not symbolized-value)
+                (eq? symbolized-value 'emacs)
+                (eq? symbolized-value 'vim))
+      (throw 'bad-emulation-made
+             (format #f "Unrecognized emulation mode: ~s" value)))
+    symbolized-value))
+
+(define-record-type*
+  <editor-preferences> editor-preferences make-editor-preferences
+  editor-preferences? this-editor-preferences
+  (background-color editor-preferences-background-color
+                    (default "000000"))
+  (text-color       editor-preferences-text-color
+                    (default "FFFFFF"))
+  (emulation-mode   editor-preferences-emulation-mode
+                    (default #f)
+                    (sanitize sanitize-emulation-mode)))
+@end lisp
+
+A user could then define their preferences like this:
+
+@lisp
+(define my-preferences
+  (editor-preferences
+    (background-color "222222")
+    (emulation-mode   'vim)))
+@end lisp
+
+The value contained in @code{my-preferences} contains a custom
+@code{background-color} and @code{emulation-mode}, but keeps the default
+@code{text-color} (@code{"FFFFFF"}). If an invalid @code{emulation-mode} had
+been specified, for example if the user passed in @code{"vi"} instead of
+@code{"vim"}, @code{sanitize-emulation-mode} would immediately throw an error.
+
+The program can access values like this:
+
+@lisp
+(editor-preferences-background-color my-preferences)
+@result{} "222222"
+(editor-preferences-text-color my-preferences)
+@result{} "FFFFFF"
+(editor-preferences-emulation-mode my-preferences)
+@result{} 'vim
+@end lisp
+
+There is no way to define setters (all instances are immutable).
+
+@node Record Inheritance
+@subsection Record Inheritance
+It is also possible to inherit from previously defined instances when creating
+new ones. Continuing with the editor example, someone might want to base their
+preferences on their friend's preferences but customize a value:
+
+@lisp
+(define friends-preferences
+  (editor-preferences
+    (inherit my-preferences)
+    (emulation-mode 'emacs)))
+@end lisp
+
+This keeps the same @code{background-color} and @code{text-color} that are
+contained in @code{my-preferences} but changes the @code{emulation-mode} to
+be @code{'emacs} instead of @code{'vim}.
+
+Sometimes it does not make sense for a field to be inherited. Suppose that the
+@code{<editor-preferences>} type is updated to contain a username so that a
+friendly greeting can be displayed when the program starts up:
+
+@lisp
+;; Usernames must be strings. It would be strange to pass a username as a
+;; symbol, so throw an error in case the user meant to pass in a variable's
+;; value instead of a literal symbol.
+(define (sanitize-username value)
+  (unless (string? value)
+    (throw 'bad-username
+           (format #f "Usernames must be strings! Got: ~s" value)))
+  value)
+
+(define (sanitize-emulation-mode value)
+  (let ((symbolized-value (cond ((not value)     #f)
+                                ((string? value) (string->symbol value))
+                                (#t              value))))
+    (unless (or (not symbolized-value)
+                (eq? symbolized-value 'emacs)
+                (eq? symbolized-value 'vim))
+      (throw 'bad-emulation-made
+             (format #f "Unrecognized emulation mode: ~s" value)))
+    symbolized-value))
+
+(define-record-type*
+  <editor-preferences> editor-preferences make-editor-preferences
+  editor-preferences? this-editor-preferences
+  (username         editor-preferences-username
+                    (innate)
+                    (sanitize sanitize-username))
+  (background-color editor-preferences-background-color
+                    (default "000000"))
+  (text-color       editor-preferences-text-color
+                    (default "FFFFFF"))
+  (emulation-mode   editor-preferences-emulation-mode
+                    (default #f)
+                    (sanitize sanitize-emulation-mode)))
+@end lisp
+
+There are a couple of differences in the new @code{username} field compared to
+the fields we looked at earlier. It is marked as @code{innate}, which means
+that it will not be inherited. For example, consider what would happen if we
+tried to define new instances like this:
+
+@lisp
+(define my-preferences
+  (editor-preferences
+    (username         "my-username")
+    (background-color "222222")
+    (emulation-mode   'vim)))
+
+(define friends-preferences
+  (editor-preferences
+    (inherit my-preferences)
+    (emulation-mode 'emacs)))
+@end lisp
+
+While the @code{friends-preferences} instance still inherits the values for
+@code{background-color} and @code{text-color}, it will not inherit the value
+for @code{username}. Furthermore, as the @code{username} field does not define
+a default value the attempted creation of @code{friends-preferences} will
+actually throw an error. Instead, we could do this:
+
+@lisp
+(define my-preferences
+  (editor-preferences
+    (username         "my-username")
+    (background-color "222222")
+    (emulation-mode   'vim)))
+
+(define friends-preferences
+  (editor-preferences
+    (inherit        my-preferences)
+    (username       "friends-username")
+    (emulation-mode 'emacs)))
+@end lisp
+
+@node @code{define-record-type*} Reference
+@subsection @code{define-record-type*} Reference
+@defmac define-record-type* name syntactic-constructor constructor predicate this-identifier fields ...
+
+Define a new record type and associated helpers.
+
+@table @var
+@item name
+A symbol used to name the type, as would normally be provided to a plain
+@code{define-record-type} form. For example, @code{<package>}.
+
+@item syntactic-constructor
+A symbol that will be used to define the user-facing constructor. For example,
+the symbol @code{package} is the syntactic constructor for the @code{<package>}
+structure.
+
+@item constructor
+A symbol that will be used to define the traditional constructor. It is used in
+the implementation of the syntactic constructor, but will not typically be used
+elsewhere. The traditional @code{make-name} (for example, @code{make-package})
+is a fine value to use here.
+
+@item predicate
+A symbol that will be used to test if a value is an instance of this record.
+For example, @code{package?}.
+
+@item this-identifier
+This symbol can be used when defining fields that need to refer to the struct
+that contains them. For an example of this, see the @code{thunked} field
+property, below.
+
+@item fields
+A set of field specifiers which take the following form:
+
+@lisp
+(field-name field-getter properties ...)
+@end lisp
+
+Each of the properties must have one of the following forms:
+
+@table @code
+@item (default @var{value})
+Defines the default value for the field, if the user does not specify one using
+the syntactic constructor.
+
+@item (innate)
+Fields marked as innate will not be inherited from parent objects (see
+Instantiating Records, below, for details of object inheritance).
+
+@item (sanitize @var{proc})
+The value given by the user will be passed into @var{proc} before being stored
+in the object. For example, consider this struct definition:
+
+@lisp
+(define-record-type* <thing> thing make-thing
+  thing?
+  this-thing
+  (name  thing-name
+         (sanitize (lambda (value)
+                     (cond ((string? value) value)
+                           ((symbol? value) (symbol->string value))
+                           (else (throw 'bad! value)))))))
+@end lisp
+
+When creating @code{thing} instances either a string or a symbol can be
+supplied but it will always be stored as a string:
+
+@lisp
+(string? (thing-name (thing (name "some-name"))))
+@result{} #t
+(string? (thing-name (thing (name 'some-name))))
+@result{} #t
+(thing (name 1994))
+@result{} Throw to key `bad!' with args `(1994)'.
+@end lisp
+
+@item (thunked)
+Fields marked as @code{thunked} will actually compute the field's value in the
+current dynamic extent which is useful when referring to fluids in a field's
+value. Furthermore, that thunk can access the record it belongs to via the
+@code{this-identifier}. For example:
+
+@lisp
+(define-record-type* <rectangle> rectangle make-rectangle
+  rectangle?
+  this-rectangle
+  (width  rectangle-width)
+  (height rectangle-height)
+  (area rectangle-area (thunked)
+                       (default (* (rectangle-width  this-rectangle)
+                                   (rectangle-height this-rectangle)))))
+
+(define base-rectangle
+  (rectangle
+    (width  2)
+    (height 4)))
+
+(define derived-rectangle
+  (rectangle
+    (inherit base)
+    (width   6)))
+
+(rectangle-area base-rectangle)
+@result{} 8
+
+(rectangle-area derived-rectangle
+@result{} 24
+@end lisp
+
+@item (delayed)
+Fields marked as @code{delayed} are similar to @code{thunked} fields, except
+that they are effectively wrapped in a @code{(delay @dots{})} form. Note that
+delayed fields cannot use @code{this-identifier}.
+@end table
+@end table
+@end defmac
+
 @node Invoking guix repl
 @section Invoking @command{guix repl}