Comments about oreilly lex & yacc, 2nd Edition:

well

Comments about oreilly lex & yacc, 2nd Edition:

I bought the book back in 1993 and it pleases me each time a read it.

If you have a project involving yacc and lex , this is the best source I have found the last years.

However, a new edition would be nice !

Notice : Take care as in certain examples memory is dynamically allocated but not freed.

Comments about oreilly lex & yacc, 2nd Edition:

I have been refering the book for past three months which has helped me a lot in constructing a compiler(using c++,tool- vc++)as a part of my academic project. The book has helped in several stages of the process. The chapter 8(Yacc ambiguities and Conflicts) helped me in debugging several errors that i came across. I am very thankful to the author.

Now the compiler is working as a win32 console based application. Currently i am trying to integrate lex and yacc with vc++(MFC)so that my compiler will have a very good user interface.I am facing problem with this integration process as c++ code snipet in the MFC application cannot call the parser(.y files).

so i request your suggestion regarding this problem.

Thanking You Narasimha.G

Comments about oreilly lex & yacc, 2nd Edition:

The book is full of code examples that show how to do things that may be very hard to re-invent. It is listed as a reference by yacc implementer. I find this book to be very interesting and detailed.

---------

Mike

Comments about oreilly lex & yacc, 2nd Edition:

I have a copy of the 1st edition and recently downloaded the sample code.

I have to concur with Andy Wong Man Boon, above, that the sample code is pretty crummy. I downloaded the code because I'd a particular interest in the SQL grammar, and found that the sql1 example didn't compile, complaining about being unable to resolve yywrap(). After fixing this problem (either by linking with -lfl or putting in a substitute yywrap()), I then found that the executable gave:



1: syntax error at FROM

SQL parse failed



when fed SELECT foo FROM bar;!

Now, I'm no expert (otherwise I wouldn't have the book), but it seems to me that simple queries like this aren't allowed by the grammar in sql1.y. I got it to work by putting the following into the grammar:



sql: query_exp

;



How trustworthy can the code, or the book, be when this kind of simple mistake is made?

Emmet

Comments about oreilly lex & yacc, 2nd Edition:

If you plan to update some of the sample code, in addition to what is explain in the previous message, I would recommend to use :

static int add_word(int type, char *word);

static int lookup_word(char *word);

for the declaration of the functions used in the first sample.

Regards, Philippe

Feel free to visit http://perso.wanadoo.fr/philippe.baucour/

Comments about oreilly lex & yacc, 2nd Edition:

Well..it's time for a 3rd edition.

After going through Chapter 1 to 7, I found

that this is a good book on the subject "lex & yacc".

Unfortunately, the examples in the book (Chap 1 to 5) do not

compile and link properly. Even if an example happens to compiled and linked

successfully, the executable will crash when executed :(

(i.e. Example 1-7 (pg. 17))

I felt that failure in compiling and linking the examples

will discourage lex & yacc newbies from experimenting and learning lex & yacc.

I've included some fixes below to make the examples compile and link properly.

(I'm using Cygwin on W2K, flex 2.5.4 and bison 1.875)

Hopefully, these fixes will help lex & yacc newbies who is reading

this book and tinkering with the examples and exercises.

-Andy Wong Man Boon

if you want fixes for the other examples)

{15} to {18}

For ch1-05.l, add...

%{

...

int lookup_word(char *word);

int add_word(int type, char *word);

/* We do not require yyunput so do not define it to avoid

'C' compiler warning */

#define YY_NO_UNPUT

%}

%option noyywrap

%%

...

%%

...

int add_word(int type, char *word) {...}

int lookup_word(char *word) {...}

For ch1-05.y, add..

%{

#include

/* prototype for lexer function */

int yylex(void);

/* prototypes for error handling functions */

int yyerror(char *s);

%}

...

%%

...

%%

...

int main(void)

{

yyin = stdin; /* Initialise yyin ! */

...

return 0;

}

int yyerror(char *s)

{

fprintf(stderr, "yyerror: %s\n", s);

return 1; /* Recovery not attempted */

}

Compile ch1-05.l and ch1-05.y as follows:-

flex ch1-05.l

bison -y -d ch1-05.y

gcc -Wall -c lex.yy.c y.tab.c

gcc -Wall -o ch1-05 lex.yy.o y.tab.o

{24}

For Ex1-10.l, the returning values from

yylex() would be clearer if you add...

%{

#include

...

#define YY_NO_UNPUT

%}

%option noyywrap

%%

...

\n { return '\n'; }

"$" { return 0; /* end of input */ }

%%

int main(void)

{

int val;

while( (val=yylex()) > 0) {

printf("token is %d\n", val);

}

printf("token is %d\n", val);

return 0;

}

Compile Ex1-10.l as follows:-

flex Ex1-10.l

gcc -Wall lex.yy.c -o Ex1-10

{59}

For ch3-01.y, the returning value of yyparse()

would be clearer if you add to the main() the following:-

int main(void)

{

...

bRet = yyparse();

if( bRet != 0) {

printf("yyparse() error: ret %d\n", bRet);

} else {

printf("yyparse() ok: ret %d\n", bRet);

}

return 0;

}

Comments about oreilly lex & yacc, 2nd Edition:

An informative and helpful source of information. The bits regarding the differences between yacc and flex are especially useful. More information about the errors generated by flex and bison would be useful. More explanations regarding the yylex, yyparse, yyparse_param, yyerror, and other yy functions are needed. Particularly, how are these functions attached to flex and bison? Also, how might a programmer manipulate these functions to obtain different results?

Comments about oreilly lex & yacc, 2nd Edition:

I bought this book because a very short article on a magazine excited me, but

after I have read it completely, I still have a lot of questions and black holes

in the way things work between Lex & Yacc.

It is a good book, with lots of examples, but for somebody like me, with no

background experience about these two tools, sometimes it went just too quickly.

I think that some intermediate steps would do the trick.

It is a good reference, anyway.