cash reward for v4 targets

Discussion:

Terence Parr

2012-11-20 21:30:17 UTC

Hi,

If anyone with serious python chops wants to tackle a python target, let me know. Other targets possibly too. Much of runtime is similar in v4 from v3 but new parsing strategy is very different/complicated.

I'm willing to pay a nonzero sum (but less than consulting rates!), depending on target.

Ter

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Eric

2012-11-20 21:35:51 UTC

Permalink

Hi Ter,

Does an F# version, not just a target count. In other words, all of the
command line tools would be included and converted to F#.

Regards, Eric

Post by Terence Parr
Hi,
If anyone with serious python chops wants to tackle a python target, let
me know. Other targets possibly too. Much of runtime is similar in v4 from
v3 but new parsing strategy is very different/complicated.
I'm willing to pay a nonzero sum (but less than consulting rates!), depending on target.
Ter
List: http://www.antlr.org/mailman/listinfo/antlr-interest
http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Juancarlo Añez

2012-11-21 03:21:31 UTC

Permalink

Terr,

That the tip for v3.4(1?) on Github actually worked with Python as target
was something good for me.

I have the know about Python, but I lack it for the v3 internals, and much
less for v4's.

If there was a guide about how the Java target is structured, I might
consider targeting Python as part of my ongoing work.

Cheers,

--
Juancarlo *Añez*

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/ant

Terence Parr

2012-11-22 23:59:15 UTC

Permalink

Hi. It's pretty complicated and I don't have a huge amount of time to write up something at the moment. I did do a summary of the API in the new book, however.
Ter

Post by Juancarlo AÃ±ez
Terr,
That the tip for v3.4(1?) on Github actually worked with Python as target
was something good for me.
I have the know about Python, but I lack it for the v3 internals, and much
less for v4's.
If there was a guide about how the Java target is structured, I might
consider targeting Python as part of my ongoing work.
Cheers,

--
Juancarlo *Añez*
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Terence Parr

2012-11-23 00:02:00 UTC

Permalink

I should probably be more specific. Sam Harwell is doing C#, C++. Then, somebody could build: python, javascript.

Ter

Post by Terence Parr
Hi,
If anyone with serious python chops wants to tackle a python target, let me know. Other targets possibly too. Much of runtime is similar in v4 from v3 but new parsing strategy is very different/complicated.
I'm willing to pay a nonzero sum (but less than consulting rates!), depending on target.
Ter
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Florin Mateoc

2012-11-23 17:28:27 UTC

Permalink

I know it's not as popular, but if there is any interest, I could do Smalltalk

Florin

________________________________
From: Terence Parr <parrt-XeM6X/PiNgr2fBVCVOL8/***@public.gmane.org>
To: antlr-interest Antlr.Org <antlr-interest-***@public.gmane.org>
Sent: Thursday, November 22, 2012 7:02 PM
Subject: Re: [antlr-interest] cash reward for v4 targets

I should probably be more specific. Sam Harwell is doing C#, C++. Then, somebody could build: python, javascript.

Ter

Hi,
If anyone with serious python chops wants to tackle a python target, let me know. Other targets possibly too. Much of runtime is similar in v4 from v3 but new parsing strategy is very different/complicated.
I'm willing to pay a nonzero sum (but less than consulting rates!), depending on target.
Ter
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Terence Parr

2012-11-23 19:36:32 UTC

Permalink

Post by Florin Mateoc
I know it's not as popular, but if there is any interest, I could do Smalltalk

I love SmallTalk! That would be a huge amount of work, however, for not that big of a payoff.

Ter

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Florin Mateoc

2012-11-23 21:13:38 UTC

Permalink

Well, I wouldn't give up my day job :)
But I also love Smalltalk and I think it would be interesting work. Although, it would probably be prudent to ask first why do you think it would be a huge amount of work (compared to other targets like javascript or python)

Florin

________________________________
From: Terence Parr <parrt-XeM6X/PiNgr2fBVCVOL8/***@public.gmane.org>
To: Florin Mateoc <fmateoc-/***@public.gmane.org>
Cc: antlr-interest Antlr.Org <antlr-interest-***@public.gmane.org>
Sent: Friday, November 23, 2012 2:36 PM
Subject: Re: [antlr-interest] cash reward for v4 targets

Post by Florin Mateoc
I know it's not as popular, but if there is any interest, I could do Smalltalk

Terence Parr

2012-11-24 01:06:30 UTC

Permalink

All targets except for the C#, because it is so close to Java, will be a lot of work :)

Ter

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Florin Mateoc

2012-11-24 04:34:16 UTC

Permalink

Well, I plan to cheat and try to translate Java to Smalltalk as much as possible. I do have a lot of experience translating Smalltalk to Java, which is a much harder problem since it also needs type inferencing.

Florin

________________________________
From: Terence Parr <parrt-XeM6X/PiNgr2fBVCVOL8/***@public.gmane.org>
To: Florin Mateoc <fmateoc-/***@public.gmane.org>
Cc: antlr-interest Antlr.Org <antlr-interest-***@public.gmane.org>
Sent: Friday, November 23, 2012 8:06 PM
Subject: Re: [antlr-interest] cash reward for v4 targets

Jim Idle

2012-11-24 06:40:03 UTC

Permalink

I don't want to rain on anyone's parade but maintaining an ANTLR
target is a LOT of work and mist people get
bored/dissatisfied/upset/married/divorced/etc and we end up with half
finished targets or ones that fall away - JavaScript, python, ruby,
C++ in v3 for instance.

OK, so this is Parrr (pun intended) fir the course and we want anyone
and everyone to have a go. However, I think that for v4, it would be
helpful to have something akin to a commissioning or approval process
such that people could see whether a target was an experiment vs
production quality.

It could be as simple as not including a target in a code release
unless it is known to be good and up to date. The other targets you
could download and use, but that in itself would tell you not to put
too much faith in it.

Jim

Post by Florin Mateoc
Well, I wouldn't give up my day job :)
But I also love Smalltalk and I think it would be interesting work. Although, it would probably be prudent to ask first why do you think it would be a huge amount of work (compared to other targets like javascript or python)
Florin
________________________________
Sent: Friday, November 23, 2012 2:36 PM
Subject: Re: [antlr-interest] cash reward for v4 targets

Post by Florin Mateoc
I know it's not as popular, but if there is any interest, I could do Smalltalk

I love SmallTalk! That would be a huge amount of work, however, for not that big of a payoff.
Ter
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Sam Harwell

2012-11-24 17:37:26 UTC

Permalink

I'm hoping the non-Java targets can be developed and maintained independently from the core ANTLR 4 repository, with absolutely no target files included in antlr/antlr4 on GitHub. In this case, we could have a standard target distribution layout and locate things via the classpath. I'll certainly be building the C# and C++ targets from this perspective.

--
Sam Harwell
Owner, Lead Developer
http://tunnelvisionlabs.com

-----Original Message-----
From: antlr-interest-bounces-***@public.gmane.org [mailto:antlr-interest-bounces-***@public.gmane.org] On Behalf Of Jim Idle
Sent: Saturday, November 24, 2012 12:40 AM
Cc: antlr-interest Antlr.Org
Subject: Re: [antlr-interest] cash reward for v4 targets

I don't want to rain on anyone's parade but maintaining an ANTLR target is a LOT of work and mist people get bored/dissatisfied/upset/married/divorced/etc and we end up with half finished targets or ones that fall away - JavaScript, python, ruby,
C++ in v3 for instance.

OK, so this is Parrr (pun intended) fir the course and we want anyone and everyone to have a go. However, I think that for v4, it would be helpful to have something akin to a commissioning or approval process such that people could see whether a target was an experiment vs production quality.

It could be as simple as not including a target in a code release unless it is known to be good and up to date. The other targets you could download and use, but that in itself would tell you not to put too much faith in it.

Jim

Post by Florin Mateoc
I know it's not as popular, but if there is any interest, I could do Smalltalk

I love SmallTalk! That would be a huge amount of work, however, for not that big of a payoff.
Ter
List: http://www.antlr.org/mailman/listinfo/antlr-interest
http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Stephen Gaito

2012-11-24 19:43:19 UTC

Permalink

Sam, Jim, Terence, and the list,

I would *strongly* encourage separating the java and core from the
non-java as well.

I have found ANTLR v3's keeping all the runtimes in the same
distribution "package" has gotten in my way a number of times....

So Sam, I applaud your distribution design goals.

Regards,

Stephen Gaito

Post by Sam Harwell
I'm hoping the non-Java targets can be developed and maintained independently from the core ANTLR 4 repository, with absolutely no target files included in antlr/antlr4 on GitHub. In this case, we could have a standard target distribution layout and locate things via the classpath. I'll certainly be building the C# and C++ targets from this perspective.
--
Sam Harwell
Owner, Lead Developer
http://tunnelvisionlabs.com
-----Original Message-----
Sent: Saturday, November 24, 2012 12:40 AM
Cc: antlr-interest Antlr.Org
Subject: Re: [antlr-interest] cash reward for v4 targets
I don't want to rain on anyone's parade but maintaining an ANTLR target is a LOT of work and mist people get bored/dissatisfied/upset/married/divorced/etc and we end up with half finished targets or ones that fall away - JavaScript, python, ruby,
C++ in v3 for instance.
OK, so this is Parrr (pun intended) fir the course and we want anyone and everyone to have a go. However, I think that for v4, it would be helpful to have something akin to a commissioning or approval process such that people could see whether a target was an experiment vs production quality.
It could be as simple as not including a target in a code release unless it is known to be good and up to date. The other targets you could download and use, but that in itself would tell you not to put too much faith in it.
Jim

Post by Florin Mateoc
I know it's not as popular, but if there is any interest, I could do Smalltalk

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Alan Cabrera

2012-11-25 16:47:14 UTC

Permalink

What do you guys think about using the service loader for this?

http://docs.oracle.com/javase/6/docs/api/java/util/ServiceLoader.html

This way, my projects' jars don't have to have classes in the org.antlr package space.

I'll certainly be building the C# and C++ targets from this perspective.

I was looking at the Target class and it seems I must inherit from it to implement a Python target. I think that it would be a good idea to copy Target to JavaTarget and then have most of Target's methods be abstract.

This would be a good thread to wrap up before the release as this will become the official TPI, Target Provider Interface, for target implementors. To this end I'll sketch out my ideas on my cloned repo.

Regards,
Alan

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Terence Parr

2012-11-25 17:04:30 UTC

Permalink

hi. i'll reply to some of these to get discussion to antlr-discussion-/JYPxA39Uh5TLH3MbocFF+G/***@public.gmane.org
thanks,
Ter

Post by Alan Cabrera

What do you guys think about using the service loader for this?
http://docs.oracle.com/javase/6/docs/api/java/util/ServiceLoader.html
This way, my projects' jars don't have to have classes in the org.antlr package space.

I'll certainly be building the C# and C++ targets from this perspective.

I was looking at the Target class and it seems I must inherit from it to implement a Python target. I think that it would be a good idea to copy Target to JavaTarget and then have most of Target's methods be abstract.
This would be a good thread to wrap up before the release as this will become the official TPI, Target Provider Interface, for target implementors. To this end I'll sketch out my ideas on my cloned repo.
Regards,
Alan
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Alan Cabrera

2012-11-23 19:41:53 UTC

Permalink

I'm looking into fiddling w/ Python. My time is limited so I can't promise anything but I think that it would be fun.

Regards,
Alan

Post by Terence Parr
I should probably be more specific. Sam Harwell is doing C#, C++. Then, somebody could build: python, javascript.
Ter

Post by Terence Parr
Hi,
If anyone with serious python chops wants to tackle a python target, let me know. Other targets possibly too. Much of runtime is similar in v4 from v3 but new parsing strategy is very different/complicated.
I'm willing to pay a nonzero sum (but less than consulting rates!), depending on target.
Ter
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Juancarlo Añez

2012-11-28 04:09:03 UTC

Permalink

Terence,

I should probably be more specific. Sam Harwell is doing C#, C++. Then,

Post by Terence Parr
somebody could build: python, javascript.

MHO is that the bounty for Python (and other languages) should go to Sam if
he can document the process he followed for producing those
targets/runtimes.

With the right docs, other people can have take on it, and they would
probably do as a contribution.

Cheers,

--
Juancarlo *Añez*

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/ma

Alexander Herz

2012-11-28 10:17:34 UTC

Permalink

Hi,

I have a combined Lexer/Parser .g file which used to generate a .tokens
file when run through antlr3.1.1.
For some reason this file is not produced any more (the actual parser
code is output as expected).

Any ideas why this happens/how it can be fixed?

Thx,
Alex

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Jim Idle

2012-11-28 10:29:33 UTC

Permalink

I don't remember for sure but when I implemented the maven plugin sone
changes were made in this area. I don't remember the tokens file being
omitted from the generated file set, or deleted after use though - it
was changes to where it is generated and where ANTLR looks for it. I
could be wrong. Check the directory where you run ANTLR and the
directory where the grammars are located (if they are different).

Generally I use separate files. If you do that, do you get the file.
Also, can I ask why you need it?

Jim

Post by Alexander Herz
Hi,
I have a combined Lexer/Parser .g file which used to generate a .tokens
file when run through antlr3.1.1.
For some reason this file is not produced any more (the actual parser
code is output as expected).
Any ideas why this happens/how it can be fixed?
Thx,
Alex
List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Alexander Herz

2012-11-28 12:50:56 UTC

Permalink

Ah,

it's moved from the source to the target directory :)

I know you guys don't recommend it but I use it to build a lookup from
token id to string
(so if I get the mismatchedtoken exception with id=x I can lookup the
corresponding string).
Afaik, it's the only way to get strings (rather than names like ","
instead of COMMA).

Alex

Post by Jim Idle
I don't remember for sure but when I implemented the maven plugin sone
changes were made in this area. I don't remember the tokens file being
omitted from the generated file set, or deleted after use though - it
was changes to where it is generated and where ANTLR looks for it. I
could be wrong. Check the directory where you run ANTLR and the
directory where the grammars are located (if they are different).
Generally I use separate files. If you do that, do you get the file.
Also, can I ask why you need it?
Jim

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Jim Idle

2012-11-30 03:04:50 UTC

Permalink

I see. I tend to code switch statements based on the token type. It is some
work, but allows better control of error messages. But if this works for
you, I personally don't see any reason not to do it. It is unlikely that
the format of the tokens file will change in 3.x

Jim

Post by Alexander Herz
Ah,
it's moved from the source to the target directory :)
I know you guys don't recommend it but I use it to build a lookup from
token id to string
(so if I get the mismatchedtoken exception with id=x I can lookup the
corresponding string).
Afaik, it's the only way to get strings (rather than names like ","
instead of COMMA).
Alex

http://www.antlr.org/mailman/options/antlr-interest/your-email-address

Post by Jim Idle
List: http://www.antlr.org/mailman/listinfo/antlr-interest

http://www.antlr.org/mailman/options/antlr-interest/your-email-address
List: http://www.antlr.org/mailman/listinfo/antlr-interest
http://www.antlr.org/mailman/options/antlr-interest/your-email-address

List: http://www.antlr.org/mailman/listinfo/antlr-interest
Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address