Errata

"a file called ch1-01.l since it is our first example"
to
"a file called ch1-02.l since it is our second example"

and code block 2, line 1 changed

% 9781565920002 ch1-01.l
to
% 9781565920002 ch1-02.l

(still in constant-wdith bold.)

#include ""y.tab.h" /* token codes from the parser */
to
#include "y.tab.h" /* token codes from the parser */

main()
{
do {
yyparse();
} while (!feof(yyin));
}

SHOULD BE:
main()
{
while(!feof(yyin)) {
yyparse();
}
}

AUTHOR: Sometime in the past decade they changed f9781565920002 so that yyin isn't
initialized until the first call to the 9781565920002er. The downloadable code is
correct.

line compiles"

Removed the lines "int index = 1;"
Also removed the blank line after the third removed line.

#* { return COMMENT; }

"#*" now reads: "#.*"

added parentheses around the lead numeric parts to group the optional exponent
with either of the leading parts:

-?((([0-9]+)|([0-9]*.[0-9]+))([eE][-+]?[0-9]+)?)

FILE *file;
to
FILE *file = NULL;

[(-file| -f) filename]

with

(-file| -f) filename

%s COMMENT
to
%x COMMENT

char *p;
(not used, should be removed, although they don't hurt anything.)
(Same with page 75.)

action { execute | menu | quite | ignore } <name>

with

action { execute | menu | quit | ignore } [<name>]

middle of the page, line which reads:

choice action execute "/bin/sh"

was replaced with:

command action execute "/bin/sh"

Example 4-8, 6th line read:

yyval.string = yytext;

The strdup() is missing. The line now reads:

yyval.string = strdup(yytext);

if(c != '\"')

with

if(c != '\'')

WHERE Foods.type = "fruit"
should be changed to
WHERE Foods.type = 'fruit';

GRANT SELECT, UPDATE (address, telephone
ON employees TO PUBLIC
should be:
GRANT SELECT, UPDATE (address, telephone
ON employees TO PUBLIC;

GRANT ALL ON foods TO tony, dale WITH GRANT OPTION
should be:
GRANT ALL ON foods TO tony, dale WITH GRANT OPTION;

GRANT REFERENCES (flavor) ON Foods TO PUBLIC
should be:
GRANT REFERENCES (flavor) ON Foods TO PUBLIC;

ORDER BY sequence ASC
should be:
ORDER BY sequence ASC;

The code near the bottom of the page read:

column '=' scalar_exp
| column '=' NULLX
;

It now reads:

column COMPARISON scalar_exp
| column COMPARISON NULLX
;

select_statement:
SELECT opt_all_distinct selection
INTO target_commalist
table_exp
;

to

select_statement:
SELECT opt_all_distinct selection
INTO target_commalist
table_exp
|
// manually added (faisal)
SELECT opt_all_distinct selection
table_exp
;

POSIX Lexxtags query replace
with

POSIX Lex

extern int *myinputlim; /* end of data */

should be changed to

extern char *myinputlim; /* end of data */

extern int myinputlim; /* end of data */
to
extern int *myinputlim; /* end of data */

Matches

with
. Matches

note the dot (.)

The code in the middle of the page read:

start: Z
| b Z

it now reads:

start: a Z
| b Z

"generated with some versions of bison"

The function "myinput" that ends here (Example E-2) won't work when "max" is
small, because the variable "offset" never gets changed. Program line 1 on
this page:

}
return copylen;
}

now reads:

} else offset += copylen;
return copylen;
}

The correction said to replace "COMPARISON" with "=".

I don't think this change is warranted. The change of "=" to COMPARISON in
assignments would allow the following (incorrect) SQL statements:

UPDATE T1 SET C1 < NULL WHERE C1 >= 23

The "<" is wrong here and should generate a syntax error. The grammar as
corrected says its right. Please put back the "=" (or better still add a
token, say ASSIGN, and merge it with the COMPARISON processing). This would
also reverse the change on line 321.

<SQL>SQLERROR TOK(SQLERROR)

Some page breaks were affected, and those pages were reprinted as well.