bug#31962: 27.0; imenu.el constructs - names and doc

bug#31962: 27.0; imenu.el constructs - names and doc

[Drew Adams]

1. `imenu--generic-function' should not have a double-hyphen in its
name.  It is the main function in the library, and it is explicitly
referenced by multiple doc strings in the file, including that of a user
option (`imenu-generic-skip-comments-and-strings') and that of the
principal configuration variable (`imenu-generic-expression'), which is
defined per mode.  The function should be renamed to
`imenu-generic-function', and the old name should be deprecated.

2. The doc string of `imenu--generic-function' says this, with zero
description of INDEX-NAME INDEX-POSITION FUNCTION and ARGUMENTS:

  The return value is an alist of the form
   (INDEX-NAME . INDEX-POSITION)
  or
   (INDEX-NAME INDEX-POSITION FUNCTION ARGUMENTS...)
  The return value may also consist of nested index alists like:
   (INDEX-NAME . INDEX-ALIST)
  depending on PATTERNS.

Each of those pieces of the return value should be specified, not just
named.

3. Similarly, other functions and variables should be renamed to use a
single, not a double, hyphen, starting with `imenu--sort-by-name' and
`imenu--index-alist'.  I don't argue now that NO functions wnd variables
in the library should have names declaring that they are internal, but a
better judgment call is in order.

They too are referenced from other doc strings.  For example, the doc
string of `imenu-sort-function' says to look at `imenu--sort-by-name'
for an example of a sort function.  And the doc string of autoloaded
variable `imenu-create-index-function' refers to `imenu--index-alist'
for the format of the buffer index alist.

We are documenting the structure and behavior of Imenu by referring to
"internal" variables and functions.  That makes no sense - or rather, it
points out how silly was declaring these to be "internal".



In GNU Emacs 27.0.50 (build 3, x86_64-w64-mingw32)
 of 2018-03-21
Repository revision: e70d0c9e66d7a8609450b2889869d16aeb0363b5
Windowing system distributor `Microsoft Corp.', version 6.1.7601
Configured using:
 `configure --without-dbus --host=x86_64-w64-mingw32
 --without-compress-install -C 'CFLAGS=-O2 -static -g3''

bug#31962: 27.0; imenu.el constructs - names and doc

[Eli Zaretskii]

> Date: Sun, 24 Jun 2018 18:51:23 -0700 (PDT)
> From: Drew Adams <drew.adams@oracle.com>
> 
> 1. `imenu--generic-function' should not have a double-hyphen in its
> name.  It is the main function in the library, and it is explicitly
> referenced by multiple doc strings in the file

As long as this function is not autoloaded, we have no unequivocal
proof that it's the main function in the library.  A function that is
the workhorse of a library can legitimately be an internal function.

I agree that this case is somewhat borderline, but I don't have enough
time and energy to make the changes in the current system, because the
problems it creates sound very minor to me.

I made small fixes in 2 doc strings, but that's as far as I was
prepared to go in this case.

> including that of a user option
> (`imenu-generic-skip-comments-and-strings') and that of the
> principal configuration variable (`imenu-generic-expression'), which
> is defined per mode.

The doc string of imenu-generic-skip-comments-and-strings was one of
those I changed.

> 2. The doc string of `imenu--generic-function' says this, with zero
> description of INDEX-NAME INDEX-POSITION FUNCTION and ARGUMENTS:
> 
>   The return value is an alist of the form
>    (INDEX-NAME . INDEX-POSITION)
>   or
>    (INDEX-NAME INDEX-POSITION FUNCTION ARGUMENTS...)
>   The return value may also consist of nested index alists like:
>    (INDEX-NAME . INDEX-ALIST)
>   depending on PATTERNS.
> 
> Each of those pieces of the return value should be specified, not just
> named.

Not as long as this is an internal function.

> 3. Similarly, other functions and variables should be renamed to use a
> single, not a double, hyphen, starting with `imenu--sort-by-name' and
> `imenu--index-alist'.  I don't argue now that NO functions wnd variables
> in the library should have names declaring that they are internal, but a
> better judgment call is in order.

Likewise.

> We are documenting the structure and behavior of Imenu by referring to
> "internal" variables and functions.  That makes no sense - or rather, it
> points out how silly was declaring these to be "internal".

Please don't use "silly" to describe work of others, as doing so makes
the discussion much less pleasant.

bug#31962: 27.0; imenu.el constructs - names and doc

[Stefan Kangas]

Eli Zaretskii <eliz@gnu.org> writes:

>> Date: Sun, 24 Jun 2018 18:51:23 -0700 (PDT)
>> From: Drew Adams <drew.adams@oracle.com>
>>
>> 1. `imenu--generic-function' should not have a double-hyphen in its
>> name.  It is the main function in the library, and it is explicitly
>> referenced by multiple doc strings in the file
>
> As long as this function is not autoloaded, we have no unequivocal
> proof that it's the main function in the library.  A function that is
> the workhorse of a library can legitimately be an internal function.
>
> I agree that this case is somewhat borderline, but I don't have enough
> time and energy to make the changes in the current system, because the
> problems it creates sound very minor to me.
>
> I made small fixes in 2 doc strings, but that's as far as I was
> prepared to go in this case.
>
>> including that of a user option
>> (`imenu-generic-skip-comments-and-strings') and that of the
>> principal configuration variable (`imenu-generic-expression'), which
>> is defined per mode.
>
> The doc string of imenu-generic-skip-comments-and-strings was one of
> those I changed.
>
>> 2. The doc string of `imenu--generic-function' says this, with zero
>> description of INDEX-NAME INDEX-POSITION FUNCTION and ARGUMENTS:
>>
>>   The return value is an alist of the form
>>    (INDEX-NAME . INDEX-POSITION)
>>   or
>>    (INDEX-NAME INDEX-POSITION FUNCTION ARGUMENTS...)
>>   The return value may also consist of nested index alists like:
>>    (INDEX-NAME . INDEX-ALIST)
>>   depending on PATTERNS.
>>
>> Each of those pieces of the return value should be specified, not just
>> named.
>
> Not as long as this is an internal function.
>
>> 3. Similarly, other functions and variables should be renamed to use a
>> single, not a double, hyphen, starting with `imenu--sort-by-name' and
>> `imenu--index-alist'.  I don't argue now that NO functions wnd variables
>> in the library should have names declaring that they are internal, but a
>> better judgment call is in order.
>
> Likewise.

It looks like Eli made the necessary changes above, and I see nothing
more to do here.  I'm therefore closing this bug report.  Please reopen
if that's incorrect.

Best regards,
Stefan Kangas