On Wed, 3 Jun 2015, Ze'ev Atlas wrote:
> Hi PhilipI enclose here my CHARTAB8 which is the EBCDIC equivalent of
> pcre_chartable.c. I Believe that it would explain some of the
> different printouts.
> OTH, I could not find where PCRE discovers \L and rejects it, PCRE
> correctly fails in both EBCDIC and ASCII. Similarly, I could not find
> where PCRE discovers \l and supposed to reject it as well but it does
> not and looks at it as if it was the charcter 'l'. I need some
> guidance on it. It looks like the last identified bug in the code.
Very odd, because the rejection of both happens together, in line 1043
of pcre_compile.c (in the current code), which reads like this:
switch (c)
{
/* A number of Perl escapes are not handled by PCRE. We give an explicit
error. */
case CHAR_l:
case CHAR_L:
*errorcodeptr = ERR37;
break;
So it looks as if somehow CHAR_l is being mis-defined. The definitions
happen in pcre_internal.h, not in the chartables. Here we find
#define CHAR_l 'l'
for both ASCII & EBCDIC (when Unicode is not defined). The only other
possibility is that control is never reaching that code. This *does*
depend on various tables (see code around line 1013).
Aha! There is a bug in line 207 - in an EBCDIC table. The line reads
/* 90 */ 0, 0, -ESC_k, 'l', 0, ESC_n, 0, -ESC_p,
Instead of 'l', there should be 0 (as there is for the L entry below).
If you can confirm that this fixes it, I'll change the master code. (I
am still having Internet issues, so my responses might be slow.)
Philip
--
Philip Hazel
