Tcl Source Code

View Ticket
Login
Ticket UUID: 821499
Title: Tcl_AppendToObj disobeys length
Type: Bug Created on: 2003-10-10 21:38:50
Submitter: dgp Assigned to: jenglish
Subsystem: 10. Objects Severity:
Priority: 5 Medium Last modified: 2004-03-27 04:58:56
Status: Open Closed by:
Resolution: None Closed on:
Version: obsolete: 8.5a0
Description:
#include "tcl.h"

int
main(int argc, char **argv)
{
    Tcl_UniChar ch = 0;
    Tcl_Obj *objPtr = Tcl_NewUnicodeObj(&ch,0);

    Tcl_AppendToObj(objPtr,"\302\242",1);
    ch = Tcl_GetUniChar(objPtr, 0);
    if (ch != 0xC2) {
        printf("BUG!\n");
        if (ch == 0xA2) {
            printf("Tcl_AppendToObj scanned > 1 byte\n");
        }
    }
}


Compile and run this test program for evidence
that Tcl_AppendToObj does not honor its
length argument.  It will scan more than "length"
bytes if it thinks it's in the middle of a UTF-8 sequence.
User Comments:
jenglish added on 2003-10-11 07:14:29:
Logged In: YES 
user_id=68433

I think a case can be made that this is an illegal use of
Tcl_AppendToObj().  It's not clear from the documentation
whether Tcl_AppendToObj() expects UTF-8 or "Lax UTF-8" as
input.  If the former, then this isn't actually a bug, since
the input is illegal.  If the latter, then it is a bug.

hobbs added on 2004-02-07 02:37:05:
Logged In: YES 
user_id=72656

let's call this a doc bug ... Tcl_AppendToObj should say
something about "undefined behavior when invalid utf-8
sequences are passed".

msofer added on 2004-03-27 04:54:37:
Logged In: YES 
user_id=148712

Doc bug about utf-8 stuff? Better leave it to people who
know: dgp?

dgp added on 2004-03-27 04:58:56:
Logged In: YES 
user_id=80530

I'm the wrong guy for
the prescribed fix.