Differences From Artifact [91d7ab8dba]:

• File generic/tclUtf.c — part of check-in [cdffcbaec9] at 2020-04-08 12:32:16 on branch sebres-utf-prev-fix — Rewrite of Tcl_UtfPrev that restores the correct behavior (no inconsistency between Tcl_UtfPrev/Tcl_UtfToUniChar anymore), Tcl_UtfPrev strictly follows all rules of Tcl_UtfToUniChar now, also fixing all test cases now. (user: sebres size: 50144)

To Artifact [d84096c4c4]:

• File generic/tclUtf.c — part of check-in [57aa5dae72] at 2020-04-08 13:26:12 on branch sebres-utf-prev-fix — partially revert [e8bfb4c2ba884b1b], a prevention added previously in string trim functions (compare char lengths) is not needed with new version of Tcl_UtfPrev (leave other optimizations as appropriate, may be important in some obscure case not covered by current or future implementation of Tcl_UtfPrev/Tcl_UtfToUniChar) (user: sebres size: 50138)

697
698
699
700
701
702
703
704
705
706
707
708
709
710
711

	    /* traversing back reaches ascii - either known mark points to char,
	     * or unexpected case with invalid utf-8 or unsupported surrogate 
	     * starting from next char, so goto next char and stop search: */
	    look++;
	    break;
	}
	o = Tcl_UtfToUniChar(look, &ch);
	/* if we are reached start of some char, mark as known and try further */
	if (o >= i) {
	    known = i;
	}
	look--;
    }
    /* use best (marked as known) char position, or current found if unknown: */
    if (known) {

|

697
698
699
700
701
702
703
704
705
706
707
708
709
710
711

	    /* traversing back reaches ascii - either known mark points to char,
	     * or unexpected case with invalid utf-8 or unsupported surrogate 
	     * starting from next char, so goto next char and stop search: */
	    look++;
	    break;
	}
	o = Tcl_UtfToUniChar(look, &ch);
	/* if we reach start of some char, mark as known and try further */
	if (o >= i) {
	    known = i;
	}
	look--;
    }
    /* use best (marked as known) char position, or current found if unknown: */
    if (known) {