From: Stuart Henderson <stu@spacehopper.org>
Subject: Re: Adding ECMA-48 REP to wscons to fix breakage
To: tech@openbsd.org
Date: Sun, 7 Jul 2024 21:22:12 +0100

On 2024/06/20 06:35, Crystal Kolipe wrote:
> The attached patch adds the CSI b control sequence to wscons.

This seems sane and useful to me, did anyone else look at it?

I only have one nit, inline below.

> This is needed because the curses updates in the last development cycle broke
> the use of the xterm emulation which had worked fine in both 7.3 and 7.4.
> 
> Unlike the quick kludge I posted last month to make things usable, this
> version is intended to fix the problem correctly and handles multiple line
> output.
> 
> Notes:
> 
> The repeat count is limited to 65535, which matches xterm behaviour.
> 
> According to the spec, sending the repeat sequence after a non-printable
> character is undefined behaviour.  For example, considering the following
> sequence:
> 
> A^G^[[5bB
> 
> Xterm displays a missing character glyph for the five repeats of the bell
> character.
> 
> Since we don't necessarily have a missing character glyph available on wscons
> this version of the patch just repeats the last printed character.
> 
> In this case we get:
> 
> AAAAAAB
> 
> diff -ur 1/sys/dev/wscons/wsemul_vt100.c 2/sys/dev/wscons/wsemul_vt100.c
> --- sys/dev/wscons/wsemul_vt100.c	Wed Aug  2 20:20:19 2023
> +++ sys/dev/wscons/wsemul_vt100.c	Thu Jun 20 10:05:51 2024
> @@ -394,6 +394,8 @@
>  		}
>  	}
>  
> +edp->last_printed_char=instate->inchar;

   ^ whitespace, missing tab

> +
>  #ifdef HAVE_DOUBLE_WIDTH_HEIGHT
>  	WSEMULOP(rc, edp, &edp->abortstate, putchar,
>  	    (edp->emulcookie, edp->crow, edp->ccol << edp->dw, dc,
> diff -ur 1/sys/dev/wscons/wsemul_vt100_subr.c 2/sys/dev/wscons/wsemul_vt100_subr.c
> --- sys/dev/wscons/wsemul_vt100_subr.c	Sun Feb 26 15:09:53 2023
> +++ sys/dev/wscons/wsemul_vt100_subr.c	Thu Jun 20 09:48:08 2024
> @@ -43,7 +43,8 @@
>  #define VTMODE_SET 33
>  #define VTMODE_RESET 44
>  #define VTMODE_REPORT 55
> -
> +extern int wsemul_vt100_output_normal(struct wsemul_vt100_emuldata *edp,
> +    struct wsemul_inputstate *instate, int kernel);
>  /*
>   * scroll up within scrolling region
>   */
> @@ -548,6 +549,17 @@
>  				break;
>  		}
>  		break;
> +	case 'b': /* REP - repeat previously printed character */
> +		instate->inchar = edp->last_printed_char;
> +		/*
> +		 * We arbitrarily limit the repeat count to 65535 to avoid
> +		 * un-interruptable flooding of the console.  This matches
> +		 * current xterm behaviour.
> +		 */
> +		for (m=0; m < (DEF1_ARG(0) < 65535 ? DEF1_ARG(0) : 65535); m++) {
> +			wsemul_vt100_output_normal(edp, instate, 0);
> +			}
> +		break;
>  	case 'c': /* DA primary */
>  		if (ARG(0) == 0)
>  			wsdisplay_emulinput(edp->cbcookie, WSEMUL_VT_ID1,
> diff -ur 1/sys/dev/wscons/wsemul_vt100var.h 2/sys/dev/wscons/wsemul_vt100var.h
> --- sys/dev/wscons/wsemul_vt100var.h	Mon Mar  6 17:14:44 2023
> +++ sys/dev/wscons/wsemul_vt100var.h	Mon May 27 16:45:26 2024
> @@ -99,6 +99,7 @@
>  #else
>  	u_char translatebuf[1];
>  #endif
> +	uint32_t last_printed_char;
>  };
>  
>  /* some useful utility macros */
>