T-Gen -- parser generator in Smalltalk 
Author Message
 T-Gen -- parser generator in Smalltalk

Is anyone trying to support (keep current and/or improve) a VisualWorks
verion of T-Gen...  

We (Tektronix) have been using T-Gen in a VisualWorks 2.51 based image
for many years.  About a year ago (maybe longer) I got a newer release
(via email from Don Roberts in the form of a parcel) for VisualWorks
3.0.  This had some minor changes/fixes from the 2.51 release.  The
vw3.0 version of T-Gen seems to parcel in and work fine in a vw5i.2
image, although we haven't stressed it yet.  

Don Robert's email last year indicated that he didn't plan on putting
much effort into T-Gen and the Illinois archive doesn't seem to have
been updated.

So, is there someone supporting or improving T-Gen?  

PS. I noticed that someone has recently ported T-Gen to squeak
        http://www.*-*-*.com/



Wed, 26 Feb 2003 05:34:56 GMT  
 T-Gen -- parser generator in Smalltalk

Quote:
> Is anyone trying to support (keep current and/or improve) a VisualWorks
> verion of T-Gen...  

Not that I know of.

[snip]

Quote:
> Don Robert's email last year indicated that he didn't plan on putting
> much effort into T-Gen and the Illinois archive doesn't seem to have
> been updated.

> So, is there someone supporting or improving T-Gen?  

I've used the Squeak verison in a couple of projects (including the ICFP
programming contest, squeak entry; although my comrades in arms might
justly claim that my bringing in T-Gen was a curse rather than a
blessing :)). There's also a Dolphin port, I believe.

I'd love to see this made into a Camp Smalltalk project. A
cross-platform, updated parser generating system would be the bomb.
Point in fact, T-Gen, as it stands and by today's standards, is pretty
lame and rather frustrating to use. Better docs, some tutorials, and
vigorous updating are in order :) Personally, I think it would be very
cool if the core parsers of *all* the Smalltalks were T-Genned :)

Anyway, I'm interested in helping, if anyone gets some momentum going.
(Maybe I'll put it on the Camp Smalltalk Wiki, if no one beats me to
it.)

--
Bijan Parsia
http://monkeyfist.com/
...among many things.



Wed, 18 Jun 1902 08:00:00 GMT  
 T-Gen -- parser generator in Smalltalk
'Point in fact, T-Gen, as it stands and by today's standards, is pretty
lame and rather frustrating to use. '

Can u expand on that a little more?  Thanks.

--tc

Sent via Deja.com http://www.deja.com/
Before you buy.



Thu, 06 Mar 2003 08:58:35 GMT  
 T-Gen -- parser generator in Smalltalk

Quote:

> 'Point in fact, T-Gen, as it stands and by today's standards, is pretty
> lame and rather frustrating to use. '

> Can u expand on that a little more?  Thanks.

For starters, the fact that the manual talks about running the thing
by clicking the middle mouse button. I am embarrassed
at how long it took me to stumble upon how to do it when
you only have a 2-button mouse.

Also, that one cannot prioritize (control firing order) of the
transformations. There are many practical situations where
the rules *are* ambigous and this ordering is required.



Thu, 06 Mar 2003 10:22:18 GMT  
 T-Gen -- parser generator in Smalltalk
My summer intern was the poor soul who had to battle it, but two things I
recall were that we couldn't get the scanner to handle double quotes the
way we wanted to, and the copyright symbol.  Also, it was too slow to use
at run-time, so we had to parse the files, dump them using the Object
Swapper, and read in the object files at run-time.  We're using the VAST
version.
Quote:

> 'Point in fact, T-Gen, as it stands and by today's standards, is pretty
> lame and rather frustrating to use. '

> Can u expand on that a little more?  Thanks.

> --tc

> Sent via Deja.com http://www.deja.com/
> Before you buy.



Wed, 18 Jun 1902 08:00:00 GMT  
 T-Gen -- parser generator in Smalltalk

Quote:

>'Point in fact, T-Gen, as it stands and by today's standards, is pretty
>lame and rather frustrating to use. '

>Can u expand on that a little more?  Thanks.

It's a bit inefficient & slow.

Ian



Wed, 18 Jun 1902 08:00:00 GMT  
 T-Gen -- parser generator in Smalltalk
Could someone please post a link to vw 3 version of T-Gen ? I couldn't find
any

Thanks

Boris


Quote:
> Is anyone trying to support (keep current and/or improve) a VisualWorks
> verion of T-Gen...

> We (Tektronix) have been using T-Gen in a VisualWorks 2.51 based image
> for many years.  About a year ago (maybe longer) I got a newer release
> (via email from Don Roberts in the form of a parcel) for VisualWorks
> 3.0.  This had some minor changes/fixes from the 2.51 release.  The
> vw3.0 version of T-Gen seems to parcel in and work fine in a vw5i.2
> image, although we haven't stressed it yet.

> Don Robert's email last year indicated that he didn't plan on putting
> much effort into T-Gen and the Illinois archive doesn't seem to have
> been updated.

> So, is there someone supporting or improving T-Gen?

> PS. I noticed that someone has recently ported T-Gen to squeak
> http://minnow.cc.gatech.edu/squeak/1419



Wed, 18 Jun 1902 08:00:00 GMT  
 T-Gen -- parser generator in Smalltalk
I sent you the parcels I was given to

as an attachment

Here's what I said...

Boris,

I have attached the (gzip'd) vw3.0 T-Gen parcels Don Roberts sent me in
January of 1999.  Maybe you could find a home for these vw3.0 parcels on
some smalltalk archive?  Maybe with the fix inserted?  (Don Roberts
hasn't responded to my private pleas to him ... :-( Maybe I have a bad
email address for him and/or maybe his seeing your interest might
inspire him!)

A minor bug in the vw3.0 parcels ...

a tgen grammar requires
GrammarTreeBuilder, not AbstractTreeBuilder....

here's the fix

!GrammarSpecParser methodsFor: 'private'!

treeBuilderClass
        ^GrammarTreeBuilder
! !

I recently noticed that vw5i.2 had trouble re-parcelling the
generated initialize method of a tgen scanner or parser.  The initialize
method sets the prodTable to be an array of literals.  The literal used
for representing the endOfInput is $x where x is the (Character value:
4), an unprintable character.  The XML format code reader in vw5i.2 does
not tolerate non-printable characters in the input.

Quote:

>     Hi
> Could you point me to the source where you got T-Gen for VisualWorks 3 ?
> Or (which is even better) send it to me ?
> I am working on a CSS parser right now and it could probably help me in my
> work.

> Thank you.

> Boris Popov.


> Could someone please post a link to vw 3 version of T-Gen ? I couldn't find
> any

> Thanks

> Boris


> > Is anyone trying to support (keep current and/or improve) a VisualWorks
> > verion of T-Gen...

> > We (Tektronix) have been using T-Gen in a VisualWorks 2.51 based image
> > for many years.  About a year ago (maybe longer) I got a newer release
> > (via email from Don Roberts in the form of a parcel) for VisualWorks
> > 3.0.  This had some minor changes/fixes from the 2.51 release.  The
> > vw3.0 version of T-Gen seems to parcel in and work fine in a vw5i.2
> > image, although we haven't stressed it yet.

> > Don Robert's email last year indicated that he didn't plan on putting
> > much effort into T-Gen and the Illinois archive doesn't seem to have
> > been updated.

> > So, is there someone supporting or improving T-Gen?

> > PS. I noticed that someone has recently ported T-Gen to squeak
> > http://minnow.cc.gatech.edu/squeak/1419



Sat, 08 Mar 2003 09:44:52 GMT  
 T-Gen -- parser generator in Smalltalk

Quote:

> 'Point in fact, T-Gen, as it stands and by today's standards, is pretty
> lame and rather frustrating to use. '

> Can u expand on that a little more?  Thanks.

Sure. Token classes can't overlap. The default token specification
grammar has no convenience constructs for various common classes of
characters (e.g., you have to do painful stuff with ~[] and alteration).
Essentially, it *really* wants you to give it an LL(1) grammar (and when
it transforms, directives get completely munged). As others have pointed
out it's rather slow. You can't do many useful pragmatic things (like
rule order). Little to no support for useful grammar debugging. No
"changes" support, so when you mess with a grammar you have to save it
explicity before experimenting. Etc. etc.

It'd be nice to have an ANTLR class tool, and perhaps something that
supported, e.g., attributative grammars.

Plus, the documentation, frankly, sucks big time.

No one's worked on it in, what, five years? It's showing some age there
:)

--
Bijan Parsia
http://monkeyfist.com/
...among many things.



Wed, 18 Jun 1902 08:00:00 GMT  
 T-Gen -- parser generator in Smalltalk
Folks interested in parsing technology might be interested
in the DMS Reengineering Toolkit, which contains generators
for UNICODE-capable lexers and parsers,
and for which a large number of legacy language grammars
have been defined.   [We have a draft for ANSI Smalltalk
done, but not ready for prime time].

It also has attribute grammars,  surface syntax rewrite rules,
prettyprinters, and a lot of other useful features, including
pretty decent documentation.

See http://www.semdesigns.com/Products/DMS/DMSToolkit.html.

--

Semantic Designs, Inc., www.semdesigns.com FAX 512-250-1191
12636 Research Blvd #C214, Austin, Texas 78759


Quote:

> > 'Point in fact, T-Gen, as it stands and by today's standards, is pretty
> > lame and rather frustrating to use. '

> > Can u expand on that a little more?  Thanks.

> Sure. Token classes can't overlap. The default token specification
> grammar has no convenience constructs for various common classes of
> characters (e.g., you have to do painful stuff with ~[] and alteration).
> Essentially, it *really* wants you to give it an LL(1) grammar (and when
> it transforms, directives get completely munged). As others have pointed
> out it's rather slow. You can't do many useful pragmatic things (like
> rule order). Little to no support for useful grammar debugging. No
> "changes" support, so when you mess with a grammar you have to save it
> explicity before experimenting. Etc. etc.

> It'd be nice to have an ANTLR class tool, and perhaps something that
> supported, e.g., attributative grammars.

> Plus, the documentation, frankly, sucks big time.

> No one's worked on it in, what, five years? It's showing some age there
> :)

> --
> Bijan Parsia
> http://monkeyfist.com/
> ...among many things.



Wed, 18 Jun 1902 08:00:00 GMT  
 
 [ 10 post ] 

 Relevant Pages 

1. Q: T-gen/Smalltalk parsers generator tools ?

2. Q: T-gen/Smalltalk parsers generator tools ?

3. Smalltalk Parser generator wanted (Tgen?)

4. Pointer needed to T-Gen parser generator

5. TGEN or other parser generators.

6. TGEN parser generator

7. Smalltalk parser generators

8. Parser Generator/Scanner Generator for Scheme?

9. Wanted: parser-generator/lexical-analyser-generator outputting Scheme

10. T-GEN and parsers

11. (VA3.0a) T-gen Parser

12. T-gen Parser Installation Help

 

 
Powered by phpBB® Forum Software