When I came accross ANTLR I thought it could be useful to do the migration. It took 1 day to get use to the tool. Much to my delight, I discovered that ANTLR can generate code in Python, which is a very dynamic language on which I am very productive.

First I was disappointed to discover that to migrate overloaded methods I had to do some type inference, which is not that straightforwards. But then I combined the power of my IDE to refactor overloaded methods into method with unique names and then applied my Python scripts. The results amazed and positively surprised my peers.

Now we no longer have that old annoying libary anymore.

Do yourself a favour
Jose San Leandro Armendariz Tue Jan 24, 2006 08:08
ANTLR allows you to face tasks not easy to accomplish without. Compiler construction theory is not something you can learn as easy as the kind of technologies you may already know. One thing is clear for me: there're some areas of IT which are worth spending time with, and ANTLR is one of them.


Very good tool, all in one approach
Sujeet Banerjee Sun Dec 18, 2005 03:57
This is a very good tool to pick up, even if you don't understand the esoteric texts on compilers or parsers.

We owe our current success in great measure to your work and thought...
Jonathan Malek Wed Dec 7, 2005 09:50
I picked ANTLR for a C# search project with a very large number of simultaneous users. We ran into only one issue under load (the solution to which I believe I found on this list, regarding the eventual failure in a multi-threaded environment of the ASTPair queue), which I was able to fix in source.

The performance is fantastic, and while it is true that the user-base seems very Java-centric, we were able to get a complex grammar in place in a matter of weeks with just one C# developer, and no prior experience in the arcana of parser theory.

Great parser
Rajesh Mon Dec 5, 2005 04:26
It was really amazing using the antlr tool and i have just started to use for message parsing. Really awesome tool i have seen.

ANTLR is a great!
Aleksey Sanin Fri Nov 18, 2005 00:34
We developed a complex event processing platform programmed via an easy to use SQL-based Continuous Computation Language (CCL). ANTLR did a great job for us. Thank you!

Really cool
Palaniappan Ramanathan Fri Nov 4, 2005 09:53
Its really cool. I thought the grammer for java would be very big.But I did a compiler from a subset of java and ANTLR helped me a lot. And it was pretty eay.

ANTLR 2.x is rock solid
Akhilesh Mritunjai Sun Oct 23, 2005 13:08
I can vouch for it. If you want further assurance, let me say that my career personally depends on ANTLR... and we have taken that decision after a LOT of analysis and study. ANTLR rocks!

ANTLR is backed by more than 2 *decades* of solid work by Terence. He has made a great contribution.

ANTLR is great
Sohail Sun Oct 23, 2005 12:25
I think it really requires understanding the process of writing the tools manually to be able to write antlr. Great job! I am grateful you were able to create this piece of work and I am glad I grew up in the time of antlr and not flex/yacc :D

We owe you a lot (concerning parsing and learning how to do that)
Zoran Bacic Sat Oct 1, 2005 10:32
Some of our engineers have your picture framed and officially blessed with our developer sweat and tears, not far from your two-times-life-size statue in gold, with titan-vanadium accents. Thanks a million for everything.

Congratulations on such a good job
Martin Mamo Sun Sep 25, 2005 11:10
I'm surprised something so good is free. I have just used ANTLR to generate a lexer and parser for a reasonably complex language. My work with it isn't finished, but so far I am very impressed. I'm surprised something so good is free. I can't see how I could have generated an equivalent parser by hand in the same amount of time as I have done with ANTLR.

ZG9K - my online calculator
Jeffrey M. Barber Sat Jul 30, 2005 21:52
I have been wanting to do an online CAS for sometime now. Now, that I will be teaching, I want to data-mine my students so I can maximize my coverage over problems of concern. antlr allowed me to fully specify the grunt of my language so I could rapidly get my expression trees presented online. And, I will save my students cash on an overpriced gameboy.

ANTLR is an extraordinarily capable tool!
Gerald B. Rosenberg, IP Attorney, NewTechLaw Thu Jun 23, 2005 16:06
My current project involves digging out and recognizing a variety of unpredictably located instances of semi-structured text from within natural language documents. Hardly your original design target, but still ANTLR is able to bring structure and precision to this very difficult problem. A concise, declarative, and highly maintainable grammar is far preferable to the classic alternatives of Bayesian inference engines and complex finite state machines. ANTLR's flexibility and capabilities are impressive.

I have to say, a stunningly useful bit of software Terence, Nice one
Robert Hill Thu Jun 9, 2005 10:21
I just love being able to single step into the generated code and see what its doing. The value of human readable code shouldn't be underestimated

Thanks for the great tool
Chris Monsanto Tue Jun 7, 2005 11:27
I'd just like to say I love your tool and can't wait till version 3.0. I've been working on a new programming language in some way, shape or form for around 3 years now and found ANTLR when searching for alternatives to the archaic FLEX/YACC. Thanks for the great tool and I hope work continues on it for a while to come.

ANTLR is great software
Lance Gutteridge Thu May 12, 2005 10:44
I was doing a project that needed a small language (well it was small when it started - isn't that the way). I considered writing my own parser on the basis that the language was simple and I was on a deadline, and I didn't want to take the learning curve hit. (I had used ANTLR some years earlier but I had forgotten most of that). In any event I decided to go with ANTLR. That certainly was the right decision.

I found all kinds of examples on various web sites and I was able to put my language together pretty quickly. It has grown and grown. I am so glad that I decided to go with a real parser instead of some cobbled together Java code that I was thinking of. Making changes to the grammar has been so easy with ANTLR.

ANTLR has saved me from the grunt work of doing all that parsing and let me concentrate on the specific constructs that I want. Thanks Terence et al.

THANKS FOR ANTLR
Bryan Ewbank Sun Feb 6, 2005 04:34
A very powerful tool that has enabled our team to produce much more code in much less time.

I'm glad I chose to start with the best tool that is out there!
Nigel Sheridan-Smith Thu Jan 20, 2005 20:50
Having been using ANTLR for around 6-7 months now, and being able to compare it against other tools such as SableCC, I can tell you that ANTLR is fantastic and a dream to use. It is incredibly flexible and powerful, allowing you to express your grammar in many different ways and providing you with multiple opportunities to insert actions of different kinds - both on the parser itself or the Abstract Syntax Tree with tree parsers. There are plenty of existing grammars available online which can be reused, or they can be used as a starting point for your own work. Additionally, they really help in showing you some of the more intricate detail in how to achieve different things with ANTLR. If you get stuck, there are numerous people on the mailing list who are willing to give you a push in the right direction who are all experts in different aspects - such as the C++/C#/Java generators or particular grammars that have already been created.

ANTLR can take a little bit of time to learn and some of the documentation and tutorials can be either too basic or too heavy (such as the "reference manual"). However, I have found it is worth the effort of learning - if you start out small with a simple, clean grammar (without actions) and then work upwards it becomes very easy to see how it all works. Furthermore, you can examine the highly readable output code understand what is going on. With ANTLR it is easy to overcome many of the common problems with grammars - syntactic and semantic predicates can help in overcoming ambiguities and non-determinisms by testing out different alternatives in "guess mode". Debugging is a breeze with recursive decent parsers, because you simply trace into each rule (as a method). However, it can take a while to run the code through to the exact input that breaks your parser, since rules can be called repeatedly based on different input.

I found that after a while, top-down parsers are highly intuitive. Of course, when I went to switch to a bottom-up parser, it did take a little bit of time to understand the differences. For this reason, it is usually difficult for people to switch between LL and LR parsers the first time. However, doing a little bit of research about the different types of parsers pays off and within under 2-3 hours you should be up and running.

ANTLR also have many features which I haven't used yet - token stream filtering, splitting and multiplexing (for when you need to parse different parts of the file in different ways), AST tree reconstructions (for rewriting/replacing/restructuring text), etc. I can see that many of these features would be handy for different situations, and are not the sort of things that you find in most compiler toolkits.

I did find a few difficult aspects though which might take a while to get your head around - when you have a non-determinism or ambiguity to be resolved, you have to know the grammar really well in order to work out the most appropriate syntactic predicate to resolve it - particularly when the non-determinism involves an alternative that continues within the rule or potentially "exits" the rule. Being able to work out what the next token might be can be challenging in these circumstances. However, ANTLR will consume greedily, so you can at least ignore non-determinisms temporarily until they become problematic. Also, many existing grammars are designed for LR parsers and this (potentially) requires a great deal of left-factoring to reduce conflicting rules and to eliminate left recursion. Whilst this can be a little time-consuming (and confusing with grammars you don't fully comprehend) you are then able to take advantage of some of the benefits - readability of the grammar is greatly improved and complexity is reduced. Furthermore, syntactic and semantic predicates make it very simple to understand how the parser/lexer will react in different situations.

One thing I can recommend though, is to reduce the clutter of your grammar by separating the rules from the actions. I usually tab in about 50 spaces so that all the actions are aligned on the right-hand side of the screen and the rules are still readable on the left. Furthermore, I also usually have a copy of the grammar that doesn't have actions - just the rules themselves. This makes it much easier for someone else to come in and maintain/modify your work.

Currently I have been using ANTLR with the Hibernate object-relational bridge to store the parsed text files into a relational database. The best way I have found of doing this is to generate Java classes containing all the contents to match your grammar (i.e. a Abstract Syntax Tree with concrete java objects). This can be time-consuming compared to SableCC's code generation, but you can limit the number of classes to exactly what you need rather than a class for each and every rule. Using actions in the parser, you can instantiate the java classes and then fill in the details as you encounter each token. I have been using the getter and setter methods that are required by Hibernate. Then I create the XML file required by Hibernate and "save" the top level constructs/classes. Hibernate takes care of the rest! The two pieces of software work really well together.

I am really looking forward to ANTLR 3.x - from what I have read from Terence's blogs and the recent workshop, it promises to remove many of the remaining issues, such as left recursion and LL(*) parsing, multi-lingual errors and some of the performance issues that have been brought up on the list. It also sounds like it will have some very nifty features - like "channels" for parsing multiple streams of tokens from the same file.

All in all, its an amazing tool. Thanks for Terence, Ric and many others who have made it available for all to use!

A Pleasure To Use
Wolfgang Haefelinger Thu Oct 28, 2004 23:39
This tool, antlr, is kind of pleasure to use. I still remember headaches I had when trying to teach lex/yacc ASN.1 couple of years ago.

Thanks for making antlr available.
giuliano carlini Fri Oct 22, 2004 16:37
It is a VERY useful tool, head and shoulders above what I've had to use before!

Thanks for ANTLR!
Achim Westermann Wed Oct 20, 2004 12:02
Many times I was able to get quick results for tricky tasks and was able to stand out in front of colleaques and supervisors thanks to ANTLR.

ANTLR is the most flexible and powerful parser generator that I know of
Thierry Miceli, creator of Ref++ Tue Jun 22, 2004 13:35
After trying several alternatives to write or generate a C++ language parser for my C++ refactoring tool (Ref++), I discovered ANTLR. Only with ANTLR I was able to create an usable and reliable parser. Thanks a lot for this great tool.

Corey O'Brien Wed May 26, 2004 10:06
ANTLR is a most impressive and powerful parser, and very easy to use. From the moment I installed the tool until the first working prototype of my utility, it took me a little less than three hours to parse caption strings from Delphi .DFM files into an XML format. As a student who gave my compiler construction professor aneurysms in college, this strikes me as particularly impressive. I humbly believe, however, that it reflects better on the straightforward nature of the tool than on my skills. Most of all I want to put in a testimonial for Terence. What a nice fellow! I found him to be a very friendly, approachable, and helpful resource when I was trying to solve a problem with the TokenStreamRewriteEngine. It was dropping whitespace from the output because I forgot to comment out the call to $setType(Token.SKIP).

ANTLR serves the whole range of developers
Oliver Zeigermann Mon Oct 27, 2003 19:34
Due to the lack of complexity ANTLR allows the beginner to write parsers in very short time, while it gives the full power to the advanced user. That's way I chose ANTLR for my XPA XML transformation framwork. ANTLR's community is unmatched by any other I know, and antlr.org is highly informative.

VB to C# translator
Jack Porterfield, CEO Immunicode Sat Oct 25, 2003 18:19
... we feel that antlr has been a tremendous and critical help to us. Thank you.

Developed makup language parser with ANTLR on .NET
Soumen Sarkar Thu Oct 9, 2003 19:51
Developed a custom markup language (like XML) parser on .NET platform with ANTLR/C#. This parser is being used in many countries over the world. I would say ANTLR is a great product and a very vibrant supportive community. The synergy makes it a very powerful tool to solve a wide vareity of problems.

Arnar Birgisson Wed Oct 8, 2003 18:22
I just love ANTLR! I have been using lex/yacc for quite a few projects, and LL-predicated lexing opens up a whole new world of possibilities.

As I have mentioned before, I am taking (or have taken) courses in compiler design and programming language theory. Discovering ANTLR has helped in making me seriously thinking about aquiring a masters degree specializing in this field of CS.

Reading posts on the antlr-interest as well as contemplating on the problems posted there has given me just as much of an understandnig on the practical side of my studies as all the lectures combined, and seeing ANTLR in action solving compicated problems (somtimes context-dependent) makes me appreciate it greatly.

A big hand for Terence and all the others involved in ANTLR's design and implementation!

'nuff said..

Lutz Hamel, Technical Manager, Thinking Machines Corp. Sat Jun 14, 2003 09:15
It is not often that a mature discipline like the field of compiler construction is shaken by a new and fundamental contribution. However, it is already clear that the technology underlying PCCTS, pred-LL(k) language recognition, has made such a fundamental contribution manifesting itself in one of the most readable parser specifications for the C++ language in the industry. Therefore, this book is required reading for anybody considering a serious programming language implementation project.

Frank DeRemer, inventor of SLR and LALR grammars and parsers; Co-Founder and President of MetaWare Inc. Sat Jun 14, 2003 09:15

This is a book after my own heart: practical! I agree with the author:

"Programmers want to use tools
[1] that employ mechanisms they understand,
[2] that are sufficiently powerful to solve their problem,
[3] that are flexible,
[4] that automate tedious tasks, and
[5] that generate output that is easily folded into their application.

This book is intended as a reference, not as a textbook or how-to book on language translation. [regarding PCCTS and associated book, the C/C++ version of ANTLR].

Richard Stearns, Co-inventor of LL(k) and recent awardee of Turing award Sat Jun 14, 2003 09:14
Because of this book, the average programmer now has easy access to the power and flexibility of top-down parsing. The ANTLR and SORCERER tools it describes should be useful for those writing any kind of a translator, including those writing sophisticated compilers. The book will be useful as a tutorial, a reference guide, and an educational tool. I recommend it highly. [regarding PCCTS and associated book, the C/C++ version of ANTLR]

Jay Levitt, Chief Mail Systems Architect, America Online, Inc. Sat Jun 14, 2003 09:12
I'm writing an anti-spam engine using an ANTLR-parsed configuration language. This engine will run on the backbone of AOL, handling more than a thousand e-mails per second. You don't get much more production than that.

So far, the support I've seen on ANTLR, from both Terence and the ANTLR community, has far exceeded what I've seen elsewhere - both for other open-source software like GNU, for which we pay beaucoup bucks for a "support contract" from a major open-source vendor, and proprietary software such as operating systems from vendors with which we have many-million-dollar relationships.

I haven't seen a problem during development that I couldn't solve within 24 hours. And, as others have echoed, most problems with ANTLR do occur at the development stage; if your grammar gets past the parser generator, it should run just fine.