Re: [fonc] success building maru in maru - a bunch of bugs I had to fix to make it happen

[Tom Koenig]

thanks   On 08/31/12, Loup Vaillant wrote:  From the maru page http://piumarta.com/software/maru/The 2.2 tarball seems to be here:http://piumarta.com/software/maru/maru-2.2.tar.gz(at the bottom of the page).However, we can see Ian did further work: maru is now at version 2.4Loup.Tom Koenig a écrit :>   Ian, where do I find the maru-2.2 tarball? Thanks Tom> On 08/19/12, Ian Piumarta wrote:> Hi Shawn,>>  > My starting point was the published sources for Maru 2.1 here:>  > http://piumarta.com/software/maru>> Thanks for posting this.>> Your issues with %typecheck,  and  were due to> generating the evaluator with %define-accessors set to> %define-safe-accessors in boot.l. Setting %define-accessors to> %define-unsafe-accessors fixes them.>> Your other observations arising from bit rot were spot-on and I've> updated the sources and made a maru-2.2 tarball that can generate a> working evaluator from emit.l and eval.l. I took the liberty of writing> subr_read from scratch, to match the one in eval.c, and updated various> other things to better match the current features.>> Although 'make test' and 'make test2' are working again, the languages> implemented by eval.c and eval.l are slightly different due to> development in the former that has not been carried forward into the latter.>> Regards,> Ian>>> ___> fonc mailing list> fonc@vpri.org > http://vpri.org/mailman/listinfo/fonc>>> ___> fonc mailing list> fonc@vpri.org> http://vpri.org/mailman/listinfo/fonc>___fonc mailing listfonc@vpri.orghttp://vpri.org/mailman/listinfo/fonc
___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc

Re: [fonc] success building maru in maru - a bunch of bugs I had to fix to make it happen

[Loup Vaillant]

From the maru page http://piumarta.com/software/maru/
The 2.2 tarball seems to be here:
http://piumarta.com/software/maru/maru-2.2.tar.gz
(at the bottom of the page).

However, we can see Ian did further work: maru is now at version 2.4

Loup.



Tom Koenig a écrit :

  Ian, where do I find the maru-2.2 tarball? Thanks Tom
On 08/19/12, Ian Piumarta wrote:
Hi Shawn,

 > My starting point was the published sources for Maru 2.1 here:
 > http://piumarta.com/software/maru

Thanks for posting this.

Your issues with %typecheck,  and  were due to
generating the evaluator with %define-accessors set to
%define-safe-accessors in boot.l. Setting %define-accessors to
%define-unsafe-accessors fixes them.

Your other observations arising from bit rot were spot-on and I've
updated the sources and made a maru-2.2 tarball that can generate a
working evaluator from emit.l and eval.l. I took the liberty of writing
subr_read from scratch, to match the one in eval.c, and updated various
other things to better match the current features.

Although 'make test' and 'make test2' are working again, the languages
implemented by eval.c and eval.l are slightly different due to
development in the former that has not been carried forward into the latter.

Regards,
Ian


___
fonc mailing list
fonc@vpri.org 
http://vpri.org/mailman/listinfo/fonc


___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc



___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc

Re: [fonc] success building maru in maru - a bunch of bugs I had to fix to make it happen

[Tom Koenig]

 Ian, where do I find the maru-2.2 tarball? Thanks Tom  On 08/19/12, Ian Piumarta wrote: Hi Shawn,> My starting point was the published sources for Maru 2.1 here:> http://piumarta.com/software/maruThanks for posting this.Your issues with %typecheck,  and  were due to generating the evaluator with %define-accessors set to %define-safe-accessors in boot.l.  Setting %define-accessors to %define-unsafe-accessors fixes them.Your other observations arising from bit rot were spot-on and I've updated the sources and made a maru-2.2 tarball that can generate a working evaluator from emit.l and eval.l.  I took the liberty of writing subr_read from scratch, to match the one in eval.c, and updated various other things to better match the current features.Although 'make test' and 'make test2' are working again, the languages implemented by eval.c and eval.l are slightly different due to development in the former that has not been carried forward into the latter.Regards,Ian___fonc mailing listfonc@vpri.orghttp://vpri.org/mailman/listinfo/fonc
___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc

Re: [fonc] success building maru in maru - a bunch of bugs I had to fix to make it happen

[Casey Ransberger]

Inline!

On Aug 19, 2012, at 3:47 PM, Ian Piumarta  wrote:

> Hi Shawn,
> 
>> My starting point was the published sources for Maru 2.1 here:
>> http://piumarta.com/software/maru
> 
> Thanks for posting this.
> 
> Your issues with %typecheck,  and  were due to generating the 
> evaluator with %define-accessors set to %define-safe-accessors in boot.l.  
> Setting %define-accessors to %define-unsafe-accessors fixes them.

Yes, playing it safe ends up in the middle of the road, which isn't safe at 
all.  

> Your other observations arising from bit rot were spot-on and I've updated 
> the sources and made a maru-2.2 tarball that can generate a working evaluator 
> from emit.l and eval.l.  I took the liberty of writing subr_read from 
> scratch, to match the one in eval.c, and updated various other things to 
> better match the current features.

Fantastic! Maru is really lovely and new bits are awesomesauce.

> Although 'make test' and 'make test2' are working again, the languages 
> implemented by eval.c and eval.l are slightly different due to development in 
> the former that has not been carried forward into the latter.

That's funny and fascinating. The bootstrap is not evolving alongside the 
bootstrapped thing. What's funny is that it can still bootstrap, and is still 
metacircular, while the prototype sits around collecting dust. It's kind of 
like the smoldering mess that's left after a rocket leaves for the moon. Good 
work!

> Regards,
> Ian

I'm really pleased to see that Maru still lives. No BCPL clone? Is Maru "it" at 
this point? Feel free of course to decline to comment on this paragraph until 
the time is right. 

P.S.

You mentioned something about a generation scavenging GC for it, did you do 
that? 'Cause if you did we might have a the roots of a fast (enough) Lisp 
that's purer than Scheme. 

'Scuse the French, but holy shit. 

--Casey
___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc

Re: [fonc] success building maru in maru - a bunch of bugs I had to fix to make it happen

[Shawn Morel]

> Your issues with %typecheck,  and  were due to generating the 
> evaluator with %define-accessors set to %define-safe-accessors in boot.l.  
> Setting %define-accessors to %define-unsafe-accessors fixes them.

ahh that makes sense now :)

> I took the liberty of writing subr_read from scratch, to match the one in 
> eval.c, and updated various other things to better match the current features.

Great, I'll take a look at the reader differences

> Although 'make test' and 'make test2' are working again, the languages 
> implemented by eval.c and eval.l are slightly different due to development in 
> the former that has not been carried forward into the latter.

I've been toying around this weekend with a few other features that would be 
nice to add, for example  dlopen/dlsym and possibly playing with writing-out / 
reading-in an image. Is there a reason (other than speed) that new features 
aren't added to the sexpr-impl going forward?

thanks,
shawn
___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc

Re: [fonc] success building maru in maru - a bunch of bugs I had to fix to make it happen

[Ian Piumarta]

Hi Shawn,

> My starting point was the published sources for Maru 2.1 here:
> http://piumarta.com/software/maru

Thanks for posting this.

Your issues with %typecheck,  and  were due to generating the 
evaluator with %define-accessors set to %define-safe-accessors in boot.l.  
Setting %define-accessors to %define-unsafe-accessors fixes them.

Your other observations arising from bit rot were spot-on and I've updated the 
sources and made a maru-2.2 tarball that can generate a working evaluator from 
emit.l and eval.l.  I took the liberty of writing subr_read from scratch, to 
match the one in eval.c, and updated various other things to better match the 
current features.

Although 'make test' and 'make test2' are working again, the languages 
implemented by eval.c and eval.l are slightly different due to development in 
the former that has not been carried forward into the latter.

Regards,
Ian


___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc

[fonc] success building maru in maru - a bunch of bugs I had to fix to make it happen

[Shawn Morel]

Folks,

The first attempt at this email didn't go through because of file size limits 
on my attachments. I've instead made a gist:
https://gist.github.com/3389334

One of the files in the gist is a rundown of all the bugs I ran into and had to 
fix to get the maru-implementation of the maru evaluator working, the others 
are my current versions of boot.l, emit.l and eval.l. I've been tinkering a lot 
in trying to learn how for myself how maru works. In a bunch of cases I rewrote 
things in more of a clojure style - so the files contain a mix os Ian's lisp 
style and clojure-ish style.

cheers,
shawn

___
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc