| 1 |
On Tue, Sep 2, 2008 at 6:28 AM, Etaoin Shrdlu <shrdlu@×××××××××××××.org> wrote: |
| 2 |
> On Tuesday 2 September 2008, 07:01, Kevin O'Gorman wrote: |
| 3 |
> |
| 4 |
>> I've been teaching myself scanners over the summer, thinking to teach |
| 5 |
>> my students this fall. |
| 6 |
>> |
| 7 |
>> It was hard because just about all of the examples are inadequate for |
| 8 |
>> one of these reasons: |
| 9 |
>> 1 They are pure lex, or pure yacc, not a combination. Ditto flex |
| 10 |
>> and bison. Most of such sample programs work okay, but they're not |
| 11 |
>> what's needed. Getting the parser and scanner to work together is the |
| 12 |
>> key thing. |
| 13 |
>> 2 They are spotty in their coverage of features |
| 14 |
>> 3 They are incomplete and fail to compile as presented; the |
| 15 |
>> documentation is very old-style UNIX in general, and very hard to |
| 16 |
>> learn the necessary fixes from it. |
| 17 |
>> 4 They are old and fail on modern versions of the software |
| 18 |
>> 5 They are just plain broken |
| 19 |
>> |
| 20 |
>> This applies to examples in the texinfo pages, the 2003 O'Reilly book, |
| 21 |
>> and online manuals. |
| 22 |
>> |
| 23 |
>> This is a very serious problem for learning these tools in their |
| 24 |
>> combined form. I'm relatively good at dealing with such stuff (having |
| 25 |
>> decades of practice), but I would not expect my undergrads to be able |
| 26 |
>> to learn with these materials. |
| 27 |
>> |
| 28 |
>> I was finally able to cobble together a working flex/bison parser from |
| 29 |
>> the bison-bridge example in an appendix to the flex info page. I'd |
| 30 |
>> like to share it, and perhaps other _working_ sample programs to the |
| 31 |
>> web at large. Does anyone know of a better venue for this than just |
| 32 |
>> some random page on my school's web server? I would like other people |
| 33 |
>> to be able to find this stuff and perhaps add to it or otherwise |
| 34 |
>> improve on it. |
| 35 |
> |
| 36 |
> When I had to study the subject, I found these documents (in addition to |
| 37 |
> the dragon book, which was of course the primary source of information |
| 38 |
> for the underlying theory): |
| 39 |
> |
| 40 |
> http://epaperpress.com/lexandyacc/ |
| 41 |
> http://www.kiv.zcu.cz/~lobaz/fjp/yacc/compiler_flex_bison.pdf |
| 42 |
> http://www.ibm.com/developerworks/linux/library/l-lexyac.html |
| 43 |
> |
| 44 |
> The above documents all contain examples of using lex+yacc together to |
| 45 |
> build an interpreter for simple languages, which is enough for an |
| 46 |
> undergrad course imho. Also, I seem to remember that the flex and bison |
| 47 |
> online manuals and the examples therein were quite accurate (but I might |
| 48 |
> be wrong, it was some time ago). |
| 49 |
> |
| 50 |
> I think a good place to discuss lex/yacc issues and get pointers to |
| 51 |
> resources is the comp.compilers newsgroup, see |
| 52 |
> http://compilers.iecc.com/index.phtml for more information, especially |
| 53 |
> the FAQ section. |
| 54 |
> |
| 55 |
> Another good place (maybe even better, since it's targeted to beginners) |
| 56 |
> is the compilers 101 mailing list, see |
| 57 |
> http://tech.groups.yahoo.com/group/compilers101/ for more information. |
| 58 |
> |
| 59 |
> |
| 60 |
|
| 61 |
Thanks, I've created a Google Group, and added these links to a page in it. |
| 62 |
Check out http://groups.google.com/group/compilerSamples |
| 63 |
|
| 64 |
-- |
| 65 |
Kevin O'Gorman, PhD |
Archives