add_defun() can define 13 functions max.

[bolangi]

The BUGS section of the manual says up to 16 functions may be defined using add_defun(). This test code fails after 13 functions are defined.

use v5.30;
use Term::ReadLine;
my $term = Term::ReadLine->new('custom function test');
my $coderef = sub { say "hello" };
my %escape_code;
read_esc_codes();
my $i = 0;
while ( my ($func_name,$seq) = each %escape_code) {
	$term->add_defun($func_name, $coderef);
	$term->bind_keyseq($seq, $func_name);
	say ++$i;
}


sub read_esc_codes {
%escape_code = qw(

  Escape  	\\e

  Insert  	\\e[2~
  Home  	\\e[1~
  PageUp  	\\e[5~

  Delete  	\\e[3~
  End  		\\e[4~
  PageDown  \\e[6~

  Up  		\\e[A
  Left  	\\e[D
  Down  	\\e[B
  Right  	\\e[C

  F1		\\eOP
  F2		\\eOQ
  F3		\\eOR
  F4		\\eOS
  F5		\\e[15~ 
  F6		\\e[17~ 
  F7		\\e[18~ 
  F8		\\e[19~ 
  F9		\\e[20~ 
  F10		\\e[21~ 
  F11		\\e[23~ 
  F12		\\e[24~ 

  Keypad/	\\eOo
  Keypad*	\\eOj
  Keypad-   \\eOm
  Keypad+   \\eOk
  Keypad7   \\eOw
  Keypad8   \\eOx
  Keypad9   \\eOy
  Keypad4   \\eOt
  Keypad5   \\eOu
  Keypad6   \\eOv
  Keypad1   \\eOq
  Keypad2   \\eOr
  Keypad3   \\eOs
  Keypad0   \\eOp
  Keypad.   \\eOn
  KeypadEnter   \\eOM
  
);
}

[hirooih]

Thank you for your report.

3 functions are already defined by this module.

--- a/Gnu.pm
+++ b/Gnu.pm
@@ -2291,6 +2291,7 @@ GTK+ support in addition to Tk.
 L<the bug tracker on GitHub|https://github.com/hirooih/perl-trg/issues>.
 
 =item C<add_defun()> can define up to 16 functions.
+And some of them are used already.
 
 =item Some functions and variables do not have test code yet.  Your
 contribution is welcome.  See F<t/readline.t> for details.

I have not found a way to eliminate this limitation. It requires to define a function dynamically.

See Line 1638-1681 of Gnu.xs for the current implementation.
If you need to define more functions, increase the size of fntbl[].

[bolangi]

I appreciate the quick response and suggestion. I see that I can distribute a version of trg modified with say, 32 custom perl functions to support the needs of my application(*1). However since I am distributing my app via debian, it will great if trg can support 32 named functions by default, if the memory costs are acceptable. https://metacpan.org/dist/Audio-Nama/view/script/nama Thank you Joel Roth

…

On Fri, Oct 20, 2023 at 07:12:34PM -0700, hirooih wrote: Thank you for your report. 3 functions are already defined by this module. ```diff --- a/Gnu.pm +++ b/Gnu.pm @@ -2291,6 +2291,7 @@ GTK+ support in addition to Tk. L<the bug tracker on GitHub|https://github.com/hirooih/perl-trg/issues>. =item C<add_defun()> can define up to 16 functions. +And some of them are used already. =item Some functions and variables do not have test code yet. Your contribution is welcome. See F<t/readline.t> for details. ``` I have not found a way to eliminate this limitation. It requires to define a function dynamically. See Line 1638-1681 of [Gnu.xs](https://github.com/hirooih/perl-trg/blob/master/Gnu.xs#L1637) for the current implementation. If you need to define more functions, increase the size of `fntbl[]`. -- Reply to this email directly or view it on GitHub: #23 (comment) You are receiving this because you authored the thread. Message ID: ***@***.***>

-- Joel Roth

[hirooih]

I think the memory costs are acceptable in these years.

The issue is how many functions are enough. This is the first request since 1996 when I released the first version. It seems that it is the time to double the size of the fntbl[].

By the way, do you really need so many add_defun() calls? Your example above needs only one add_defun() call.
As described here a custom function has two arguments. You can use the second argument, $key, to distinguish which key sequence is used. Does this solve your issue?

[bolangi]

On Sat, Oct 21, 2023 at 07:51:53PM -0700, hirooih wrote: I think the memory costs are acceptable in these years. The issue is how many functions are enough. This is the first request since 1996 when I released the first version. It seems that it is the time to double the size of the `fntbl[]`. By the way, do you really need so many `add_defun()` calls? Your example above needs only one `add_defun()` call. As described [here](https://metacpan.org/pod/Term::ReadLine::Gnu#Custom-Functions) a custom function has two arguments. You can use the second argument, `$key`, to distinguish which key sequence is used. Does this solve your issue?

Yes, I could certainly dispatch on the key sequence.

…

-- Reply to this email directly or view it on GitHub: #23 (comment) You are receiving this because you authored the thread. Message ID: ***@***.***>

-- Joel Roth

[hirooih]

Does this solve your issue?

Yes, I could certainly dispatch on the key sequence.

Good!

After some thought, I decide not to change the size of the table this time.

I will add a link to this issue ticket in the bug section of the documentation.
I will change it when I really have a good reason to do it.

Again, thank you for your report.

[bolangi]

On Sun, Oct 22, 2023 at 06:40:26AM -0700, hirooih wrote: >> Does this solve your issue? > Yes, I could certainly dispatch on the key sequence. Good! After some thought, I decide not to change the size of the table this time. I will add a link to this issue ticket in the bug section of the documentation. I will change it when I really have a good reason to do it. Again, thank you for your report.

That's fine, but I will say it's a more complicated API to deal with to avoid bumping into the limit. Admittedly a slow worker, I've spent an hour working on something that was already working before. Thank you for your awesome software.

…

-- Reply to this email directly or view it on GitHub: #23 (comment) You are receiving this because you authored the thread. Message ID: ***@***.***>

-- Joel Roth

[bolangi]

On Sun, Oct 22, 2023 at 10:40:18AM -1000, Joel Roth wrote: On Sun, Oct 22, 2023 at 06:40:26AM -0700, hirooih wrote: > >> Does this solve your issue? > > > Yes, I could certainly dispatch on the key sequence. > > Good!

Hi, I tried to dispatch on the key sequence as you suggest. I encounter a problem that several keys return the same code. In particular all of Insert/Home/PageUp/Delete/End/PageDown return code 126, which is only the last of the four codes returned by showkey -a in my xterm. I'm attaching test code to demonstrate. Thanks for your attention. use v5.30; use Term::ReadLine; my $term = Term::ReadLine->new('custom function test'); my %escape_code; read_esc_codes(); my $func_name = 'hotkey_dispatch'; my $coderef = \&hotkey_dispatch; $term->add_defun($func_name, $coderef); my $i = 0; while ( my ($keyname,$seq) = each %escape_code) { $term->bind_keyseq($seq, $func_name); say ++$i; } our %function; sub hotkey_dispatch { say join ' ',@_ } sub read_esc_codes { %escape_code = qw( Escape \\e Insert \\e[2~ Home \\e[1~ PageUp \\e[5~ Delete \\e[3~ End \\e[4~ PageDown \\e[6~ Up \\e[A Left \\e[D Down \\e[B Right \\e[C F1 \\eOP F2 \\eOQ F3 \\eOR F4 \\eOS F5 \\e[15~ F6 \\e[17~ F7 \\e[18~ F8 \\e[19~ F9 \\e[20~ F10 \\e[21~ F11 \\e[23~ F12 \\e[24~ Keypad/ \\eOo Keypad* \\eOj Keypad- \\eOm Keypad+ \\eOk Keypad7 \\eOw Keypad8 \\eOx Keypad9 \\eOy Keypad4 \\eOt Keypad5 \\eOu Keypad6 \\eOv Keypad1 \\eOq Keypad2 \\eOr Keypad3 \\eOs Keypad0 \\eOp Keypad. \\eOn KeypadEnter \\eOM ); our %keyname = ( reverse %escape_code ); } while ( defined ($_ = $term->readline('prompt>')) ) {say "got a line: $_" }

[hirooih]

Sorry I was wrong.
The second argument $key is an int value. It does not work for a key sequence.
We have to use rl_executing_keyseq variable.

sub hotkey_dispatch {
    say string_to_hex_dump($term->Attribs->{executing_keyseq});
}
sub string_to_hex_dump {
    my ($string) = @_;
    my $hex_dump = '';
    for my $char (split //, $string) {
        $hex_dump .= sprintf("%02X ", ord($char));
    }
    return $hex_dump;
}

[bolangi]

With the help of your code I reduced to one the number of custom functions needed to dispatch from special key character combinations. Here is an improved version of my sample code. It identifies the key pressed, and shows the sequence in various formats. I'm curious that this script doesn't correctly capture many of the key combinations (such as Ctrl-Insert, Shift-F3) that showkey -a displays. use v5.30; use Term::ReadLine; my $term = Term::ReadLine->new('keyboard code and custom dispatch test'); my %escape_code; # Insert -> \e[2~ my %keyname; # \e[2~ -> Insert read_esc_codes(); my $func_name = 'hotkey_dispatch'; my $coderef = \&hotkey_dispatch; $term->add_defun($func_name, $coderef); while ( my ($keyname,$seq) = each %escape_code) { $term->bind_keyseq($seq, $func_name); } sub hotkey_dispatch { my ($seq, $decimal, $hex) = string_to_escape_code($term->Attribs->{executing_keyseq}); my $name = $keyname{$seq}; say "Special key: $name, escape sequence: $seq, decimal: $decimal, hex: $hex"; say } sub string_to_escape_code { my ($string) = @_; my $esc = ''; my $decimal = ''; my $hex = ''; for my $char (split //, $string) { my $ord = ord($char); # It seems that readline intercepts control characters # so don't try to convert them. $char = '\e' if $ord == 27; $decimal .= "$ord "; $esc .= $char; $hex .= sprintf("%02X ", $ord); } $esc, $decimal, $hex; } sub read_esc_codes { %escape_code = qw( Escape \\e Insert \\e[2~ Delete \\e[3~ Home \\e[1~ End \\e[4~ PageUp \\e[5~ PageDown \\e[6~ Up \\e[A Left \\e[D Down \\e[B Right \\e[C F1 \\eOP F2 \\eOQ F3 \\eOR F4 \\eOS F5 \\e[15~ F6 \\e[17~ F7 \\e[18~ F8 \\e[19~ F9 \\e[20~ F10 \\e[21~ F11 \\e[23~ F12 \\e[24~ Keypad/ \\eOo Keypad* \\eOj Keypad- \\eOm Keypad+ \\eOk Keypad7 \\eOw Keypad8 \\eOx Keypad9 \\eOy Keypad4 \\eOt Keypad5 \\eOu Keypad6 \\eOv Keypad1 \\eOq Keypad2 \\eOr Keypad3 \\eOs Keypad0 \\eOp Keypad. \\eOn KeypadEnter \\eOM ); %keyname = ( reverse %escape_code ); } while ( defined ($_ = $term->readline('prompt>')) ) {say "got a line: $_" }

…

__END__

-- Joel Roth

[hirooih]

I'm curious that this script doesn't correctly capture many of the key combinations (such as Ctrl-Insert, Shift-F3) that showkey -a displays.

I have never heard of the command showkey before.

On my terminal Ctrl-Insert displays nothing, Shift-F3 displays \e[1;2R (6 characters). There is no such long sequence in your example.

Does this ( rl_set_timeout ) help you?

[bolangi]

I've debugged my problems. Thank you for your assistance. You may close this issue. regards,

…

-- Joel Roth

[hirooih]

Thank you for your assistance.

You are welcome.

I've created PR #24. It will be included in the next release.

Thanks!.