Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.forth > #30943 > unrolled thread

BOOST for Forth?

Back to article view | Back to comp.lang.forth

Contents

  BOOST for Forth? Mark <marvin_littlewood_426716@yahoo.co.uk> - 2014-05-29 04:40 -0700
    Re: BOOST for Forth? Mark Wills <markwills1970@gmail.com> - 2014-05-29 06:44 -0700
      Re: BOOST for Forth? Mark <marvin_littlewood_426716@yahoo.co.uk> - 2014-05-29 07:05 -0700
    Re: BOOST for Forth? Mark Wills <markwills1970@gmail.com> - 2014-05-29 08:15 -0700
      Re: BOOST for Forth? Alexander Skobelev <al.skobelev@gmail.com> - 2014-05-29 19:34 +0400
        Re: BOOST for Forth? rickman <gnuarm@gmail.com> - 2014-05-29 13:11 -0400
          Re: BOOST for Forth? stephenXXX@mpeforth.com (Stephen Pelc) - 2014-05-29 20:06 +0000
            Re: BOOST for Forth? rickman <gnuarm@gmail.com> - 2014-05-29 21:24 -0400
            Re: BOOST for Forth? Paul E Bennett <Paul_E.Bennett@topmail.co.uk> - 2014-06-09 10:00 +0100
    Re: BOOST for Forth? mhx@iae.nl - 2014-05-29 17:41 -0700
    Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-05-29 20:44 -0700
      Re: BOOST for Forth? Mark Wills <markwills1970@gmail.com> - 2014-05-30 02:13 -0700
        Re: BOOST for Forth? Mark <marvin_littlewood_426716@yahoo.co.uk> - 2014-05-30 09:05 -0700
          Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-05-30 21:31 -0700
            Re: BOOST for Forth? Mark <marvin_littlewood_426716@yahoo.co.uk> - 2014-05-31 04:53 -0700
              Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-06-01 18:55 -0700
              Re: BOOST for Forth? Frank Thomason <Frank@example.net> - 2014-06-02 14:31 -0400
                Re: BOOST for Forth? rickman <gnuarm@gmail.com> - 2014-06-02 18:37 -0400
                  Re: BOOST for Forth? Marvin <marvin_littlewood_426716@yahoo.co.uk> - 2014-06-02 19:12 -0700
                    Re: BOOST for Forth? rickman <gnuarm@gmail.com> - 2014-06-02 22:50 -0400
                      Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-06-03 20:54 -0700
                        Re: BOOST for Forth? rickman <gnuarm@gmail.com> - 2014-06-04 00:11 -0400
                          Re: BOOST for Forth? jfong <jfong@ms4.hinet.net> - 2014-06-04 20:05 -0700
                Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-06-02 20:30 -0700
                  Re: BOOST for Forth? Julian Fondren <ayrnieu@gmail.com> - 2014-06-02 22:28 -0700
                    Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-06-03 21:25 -0700
                Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-06-02 21:41 -0700
              Re: BOOST for Forth? hughaguilar96@yahoo.com - 2014-06-02 22:10 -0700
    BOOST for Forth? Julian Fondren <ayrnieu@gmail.com> - 2014-06-02 23:00 -0700
      Re: BOOST for Forth? Paul Rubin <no.email@nospam.invalid> - 2014-06-02 23:06 -0700
      Re: BOOST for Forth? Marvin <marvin_littlewood_426716@yahoo.co.uk> - 2014-06-10 14:19 -0700

Page 1 of 2  [1] 2  Next page →

#30943 — BOOST for Forth?

Hi clf

I am a relatively new user of gforth on Android. Reading posts here and elsewhere, it would appear that optional words in the ANSI standard are considered to define standard libraries of words but not their implementation. (See the discussion about PLACE and +PLACE.)

Coming from a C++ background, IMHO the C++ community has found BOOST a useful environment in which to share, peer review and standardise (where appropriate) libraries of code that help solve a diverse range of problems.

Is there an equivalent for Forth? (i.e. a set of public domain libraries that one can download and use)

If there isn't, is there interest in/need for such an environment?

If there is interest, is there the manpower/resource to nurture it?

Many thanks

Mark

[toc] | [next] | [standalone]

#30946

Hi Mark, welcome.

Have a look at FLAG (Forth Library Action Group):

http://soton.mpeforth.com/flag/

Regards

Mark (Wills)

[toc] | [prev] | [next] | [standalone]

#30947

Hi Mark

Many thanks for the kind welcome! Will do!

Mark ( :-) )

[toc] | [prev] | [next] | [standalone]

#30948

There's also the Forth Scientific Library:

http://www.taygeta.com/fsl/sciforth.html

However, it appears to have dropped off the net.
Others may have a more up-to-date link.

Cheers

[toc] | [prev] | [next] | [standalone]

#30949

Mark Wills <markwills1970@gmail.com> writes:

> There's also the Forth Scientific Library:
>
> http://www.taygeta.com/fsl/sciforth.html
>
> However, it appears to have dropped off the net.
> Others may have a more up-to-date link.

As it was earlier discussed here in some other thread,
FSL is available through FTP:
ftp://ftp.taygeta.com/pub/forth/Scientific

[toc] | [prev] | [next] | [standalone]

#30950

On 5/29/2014 11:34 AM, Alexander Skobelev wrote:
> Mark Wills <markwills1970@gmail.com> writes:
>
>> There's also the Forth Scientific Library:
>>
>> http://www.taygeta.com/fsl/sciforth.html
>>
>> However, it appears to have dropped off the net.
>> Others may have a more up-to-date link.
>
> As it was earlier discussed here in some other thread,
> FSL is available through FTP:
> ftp://ftp.taygeta.com/pub/forth/Scientific

I see links for the FSL at the FLAG site, but they are the broken 
versions.  Perhaps the FLAG site should update their FSL page.

-- 

Rick

[toc] | [prev] | [next] | [standalone]

#30959

On Thu, 29 May 2014 13:11:20 -0400, rickman <gnuarm@gmail.com> wrote:

>I see links for the FSL at the FLAG site, but they are the broken 
>versions.  Perhaps the FLAG site should update their FSL page.

Perhaps *you* will do your part and send me the correct links.

Stephen

-- 
Stephen Pelc, stephenXXX@mpeforth.com
MicroProcessor Engineering Ltd - More Real, Less Time
133 Hill Lane, Southampton SO15 5AF, England
tel: +44 (0)23 8063 1441, fax: +44 (0)23 8033 9691
web: http://www.mpeforth.com - free VFX Forth downloads

[toc] | [prev] | [next] | [standalone]

#30976

On 5/29/2014 4:06 PM, Stephen Pelc wrote:
> On Thu, 29 May 2014 13:11:20 -0400, rickman <gnuarm@gmail.com> wrote:
>
>> I see links for the FSL at the FLAG site, but they are the broken
>> versions.  Perhaps the FLAG site should update their FSL page.
>
> Perhaps *you* will do your part and send me the correct links.

???

-- 

Rick

[toc] | [prev] | [next] | [standalone]

#31318

Stephen Pelc wrote:

> On Thu, 29 May 2014 13:11:20 -0400, rickman <gnuarm@gmail.com> wrote:
> 
>>I see links for the FSL at the FLAG site, but they are the broken
>>versions.  Perhaps the FLAG site should update their FSL page.
> 
> Perhaps *you* will do your part and send me the correct links.
> 
> Stephen
> 

This one seems to get somewhere for FSL.

<http://c2.com/cgi/wiki?ForthScientificLibrary>

or:-

<ftp://ftp.taygeta.com/pub/Forth/Scientific/>

-- 
********************************************************************
Paul E. Bennett IEng MIET.....<email://Paul_E.Bennett@topmail.co.uk>
Forth based HIDECS Consultancy.............<http://www.hidecs.co.uk>
Mob: +44 (0)7811-639972
Tel: +44 (0)1235-510979
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************

[toc] | [prev] | [next] | [standalone]

#30973

On Thursday, May 29, 2014 1:40:20 PM UTC+2, Mark wrote:
[..]
> Is there an equivalent for Forth? (i.e. a set of public domain 
> libraries that one can download and use)
> 
See also The Forth Foundation Library (FFL). Rev. 0.9 is from 2011.
It runs on most Forths.

-marcel

[toc] | [prev] | [next] | [standalone]

#30994

On Thursday, May 29, 2014 4:40:20 AM UTC-7, Mark wrote:
> Is there an equivalent for Forth? (i.e. a set of public domain libraries that one can download and use)

See my novice package:
http://www.forth.org/novice.html

[toc] | [prev] | [next] | [standalone]

#31011

Despite the bad rap that Hugh's Novice Library has taken over the years, I like it. I have not had occasion to use the routines in the library for my own needs, but I have looked at the code, and I like the way it is written, and it is definately useful for a novice Forther to see how something might be implemented in ANS Forth. It's been useful for me in that resepct.

[toc] | [prev] | [next] | [standalone]

#31031

On Friday, 30 May 2014 10:13:10 UTC+1, Mark Wills  wrote:
> Despite the bad rap that Hugh's Novice Library has taken over the years, I like it. [...] it is definately useful for a novice Forther to see how something might be implemented in ANS Forth. It's been useful for me in that resepct.

Interesting. As an educator, I teach my students to write for literacy: can they or other humans read their code?

I have been looking at the implementation files in gforth and at Stephen Pelc's book (online version). For me, as has been written elsewhere in clf, the biggest immediate issue in my (re)learning Forth is disentangling the stack manipulations.

For example, in Stephen Pelc's book the diary application uses 2 apparently different arrays. To me, they are an excellent teaching example of refactoring because the two arrays are in fact variants of the same underlying theme. They share common operations, which are not immediately apparent as a result of the stack manipulations. I am ashamed to say that knowledge of the C++ standard library helped me see the wood for the trees!

Another example in the gforth manual was instructive. The word to find the length of a list when written using locals took 19 steps but was easy to understand. The version using the stack took 11 steps but seemed a little opaque.

Still, I haven't had as much fun writing code in ages. Compared with C++, Forth offers great freedom:

1) not much to write before you can see what happens (no header and implementation files)
2) a quick write test write cycle (no waiting seconds to minutes for code to compile and no immediate need to write formal tests to make the newest bit of code run)

So congratulations to Bernd Paysan and colleagues for making gforth available on Android.

Mark

[toc] | [prev] | [next] | [standalone]

#31067

On Friday, May 30, 2014 9:05:00 AM UTC-7, Mark wrote:
> On Friday, 30 May 2014 10:13:10 UTC+1, Mark Wills  wrote:
> 
> > Despite the bad rap that Hugh's Novice Library has taken over the years, I like it. [...] it is definately useful for a novice Forther to see how something might be implemented in ANS Forth. It's been useful for me in that resepct.

Thanks for the kind words Mark! :-)

Previously, the most positive thing that anybody had said about my novice package was Paul Rubin who said that it was a lot of tedious code that anybody could have written. You now have hit the new high mark though. 

Paul Rubin's comment was actually true though. When I originally wrote it, I saw it as being pretty basic stuff (hence the name) that I was providing to the novice who was unprepared to write it himself. 1ARRAY etc. were tedious to write, but they do provide an efficient implementation (most novices would use CREATE DOES> that is hugely inefficient). Also, and more importantly, the novice package provides all of these basic data-structures ready to go --- the programmer doesn't have to write everything from scratch every time --- it is tedious to reinvent the wheel over and over, and doing so usually results in having the same data-structure in different programs or in different parts of the same program, implemented with a slightly different technique each time.

The newcomer Mark should be made aware that Elizabeth Rather has said that I'm a failure as a Forth programmer, and will never learn Forth programming. If he should make use of my novice package in his Forth programming, she will say the same thing about him. Also, she says that it is impossible to write a general-purpose data-structure, and that it is necessary to write every data-structure from scratch for every program --- totally contradicting the basic premise of my novice package.

> For example, in Stephen Pelc's book the diary application uses 2 apparently different arrays. To me, they are an excellent teaching example of refactoring because the two arrays are in fact variants of the same underlying theme. They share common operations, which are not immediately apparent as a result of the stack manipulations. I am ashamed to say that knowledge of the C++ standard library helped me see the wood for the trees!

I can't comment on Pelc's book as I haven't read it. This sounds like what I was describing above though --- implementing arrays from scratch every time they are needed, rather than having a common general-purpose array available that can be used in any program.

> Another example in the gforth manual was instructive. The word to find the length of a list when written using locals took 19 steps but was easy to understand. The version using the stack took 11 steps but seemed a little opaque.

I use linked lists as my primary data-structure. I was inspired by Factor that uses "sequences" (resizable arrays) throughout --- I think that it is a good idea to have one data-structure used consistently throughout. Lisp uses lists --- the Lispers have other data structures (such as LLRB trees) available --- but they are encouraged to use lists if doing so is reasonable.

The following is my code (in LIST.4TH) for finding the length of a list (words you don't recognize such as W and FIELD are defined in NOVICE.4TH). Whether it is opaque or clear, I don't know. If you want examples of Forth style, you should look at my example programs --- the low-level support code (such as LENGTH or 1ARRAY etc.) was written primarily for efficiency, with readability being a secondary consideration.

SwiftForth? [if]
    
icode length ( head -- count )    
    zero(eax)
    begin  non-zero? while  [@]  eax inc  repeat
    eax ebx mov  ret end-code
    
[else]    
    
0
    w field .fore   
constant list

: length ( head -- count )
    0 0 do
        dup 0= if  I  nip  unloop exit then
        .fore @  loop ;

[then]

[toc] | [prev] | [next] | [standalone]

#31071

On Saturday, 31 May 2014 05:31:25 UTC+1, hughag...@yahoo.com  wrote:
> [...] it is tedious to reinvent the wheel over and over, and doing so usually results in having the same data-structure in different programs or in different parts of the same program, implemented with a slightly different technique each time.

Agreed. The C++ standard and the BOOST libraries avoid this and can be relied upon to work because the developers are responsive to feedback and testing using supported compilers on supported architectures.

> The newcomer Mark should be made aware that Elizabeth Rather has said that I'm a failure as a Forth programmer, and will never learn Forth programming. 

Ouch!

> [...] it is impossible to write a general-purpose data-structure, and that it is necessary to write every data-structure from scratch for every program 

This echo's a comment of Moore's quoted in Brodie's "Thinking Forth". The success of BOOST and the C++ standard library suggest otherwise. The rationale behind their design is open to scrutiny. This allows programmers to decide when it is worth "going it alone".

My programming time is limited. I am not keen on writing yet another array, as a prelude to getting on with the problem I really want to solve. I would rather have a ready made array, that I can extend if the evidence requires me to.

> SwiftForth? [if]
> icode length ( head -- count )    
>     zero(eax)
>     begin  non-zero? while  [@]  eax inc  repeat
>     eax ebx mov  ret end-code
> [else] [...] [then]

As an exemplar of the intent behind my original question, I would like to ask Hugh a couple of questions about the above code, that he very kindly supplied:

1) Is SwiftForth only offered on computer architectures where "eax" and "ebx" have meaning?
2) Is there evidence that this register optimization truly results in a significant increase in speed?

Mark (S) - i.e. not Wills

[toc] | [prev] | [next] | [standalone]

#31143

On Saturday, May 31, 2014 4:53:51 AM UTC-7, Marvin wrote:
> My programming time is limited. I am not keen on writing yet another array, as a prelude to getting on with the problem I really want to solve. I would rather have a ready made array, that I can extend if the evidence requires me to.

Well, it seems that we are pretty much in agreement. The best way to learn Forth programming, is to write Forth programs. I think that novices tend to get bogged down in implementing data-structures such as arrays and lists, etc. that they aren't prepared to implement, and they don't get their program written. They lose interest in Forth and drop the subject without ever written a program that did anything. With my novice package, they can get going on an application program --- if they get their program written they will have a sense of accomplishment, and they will feel inspired to write bigger and better programs --- there is no sense of accomplishment for having implemented a linked-list that was never used for anything.
> 
> > SwiftForth? [if]
> > icode length ( head -- count )    
> >     zero(eax)
> >     begin  non-zero? while  [@]  eax inc  repeat
> >     eax ebx mov  ret end-code
> > [else] [...] [then]
> 
> As an exemplar of the intent behind my original question, I would like to ask Hugh a couple of questions about the above code, that he very kindly supplied:
> 
> 1) Is SwiftForth only offered on computer architectures where "eax" and "ebx" have meaning?

My code was tested on SwiftForth version-2 for Windows. AFAIK, it works on version-3 as well. There may be SwiftForth versions for other processors such as the ARM, and it won't work there. If it doesn't work, then just turn the SwiftForth? compiler directive off --- it is off by default --- when it is off, the novice package should work on any ANS-Forth compiler (it has been tested on SwiftForth, Gforth and Win32Forth).

> 2) Is there evidence that this register optimization truly results in a significant increase in speed?

Yes. SwiftForth does very little optimization, and generates very inefficient code. You can see this for yourself using the SEE command. Also, I have bench-marked large programs of my own under SwiftForth with the SwiftForth? compiler directive turned off and on, and they are typically about twice as fast with SwiftForth? turned on. The functions that have an assembly-language version (such as LENGTH) are typically an order of magnitude faster, but the program's speed increase depends upon how often these functions are being used. If you use lists a lot, which I recommend, then you should get a pretty good boost in speed because I have a lot of low-level list functions written in assembly-language.

VFX does a lot more optimization, so hand-writing functions in assembly-language isn't going to help very much, as the result will be pretty much the same code either way. SwiftForth needs a lot of help --- even with a heavy dose of hand-written assembly-language, it will typically be slower than VFX. I haven't actually tested my novice package under VFX, but I have written some small test programs and used SEE to find out what is getting generated. I'll test the novice package under VFX for my next release of the novice package.

[toc] | [prev] | [next] | [standalone]

#31153

Mark,

Just for a little perspective, Elizabeth Rather has never referred to 
Hugh as a failure as a Forth programmer. She is/was a professional who 
would not speak of anyone in those terms. A few years ago, someone used 
the term "sucks" in reference to some of Hugh's work. If I recall 
correctly, however, the opinion was referring to Hugh's choice of 
algorithm, not his code in general. Since then, Hugh has increasingly 
blown this out of proportion by claiming that "everybody" says his code 
sucks, etc.

Elizabeth also never said that it is "impossible" to write general 
purpose data structures. Her statements have been to the effect that 
most applications have specific data structure needs, and that it is 
more efficient to use the power of Forth to create application-specific 
structures than to spend time writing and debugging elaborate general 
purpose structures that would be over-kill for most applications.

Please note that I have nothing against Hugh as a programmer; I have 
never had a need to review any of his code. I am just one of many people 
on this newsgroup that are getting very tired of his constant unprovoked 
attacks on Ms. Rather.

[toc] | [prev] | [next] | [standalone]

#31154

On 6/2/2014 2:31 PM, Frank Thomason wrote:
> Mark,
>
> Just for a little perspective, Elizabeth Rather has never referred to
> Hugh as a failure as a Forth programmer. She is/was a professional who
> would not speak of anyone in those terms. A few years ago, someone used
> the term "sucks" in reference to some of Hugh's work. If I recall
> correctly, however, the opinion was referring to Hugh's choice of
> algorithm, not his code in general. Since then, Hugh has increasingly
> blown this out of proportion by claiming that "everybody" says his code
> sucks, etc.
>
> Elizabeth also never said that it is "impossible" to write general
> purpose data structures. Her statements have been to the effect that
> most applications have specific data structure needs, and that it is
> more efficient to use the power of Forth to create application-specific
> structures than to spend time writing and debugging elaborate general
> purpose structures that would be over-kill for most applications.
>
> Please note that I have nothing against Hugh as a programmer; I have
> never had a need to review any of his code. I am just one of many people
> on this newsgroup that are getting very tired of his constant unprovoked
> attacks on Ms. Rather.

Just to give some weight to Frank's comments.  Hugh posts here a lot. 
Sometimes his posts are useful showing his experience with Forth.  Other 
times his posts are long winded rants displaying paranoia and possibly 
delusions, certainly displaying an extreme misunderstanding of social 
interaction.

I feel in general it is better if the latter type of posts or comments 
within a post are not responded to in any way.  Acknowledging such 
comments, even to refute them just leads to more of the same.

To put all this in perspective, Hugh just barely stands out in the crowd 
that is attracted to c.l.forth.  It is a very mixed bag of nuts, likely 
myself included... lol

-- 

Rick

[toc] | [prev] | [next] | [standalone]

#31158

Hi Rick (and Frank)

Your comments are well made! I like the notion of a mixed bag of Forth nuts!

I also like the fact that Hugh was willing to answer questions about his code.

Mark (S)
 

[toc] | [prev] | [next] | [standalone]

#31159

On 6/2/2014 10:12 PM, Marvin wrote:
> Hi Rick (and Frank)
>
> Your comments are well made! I like the notion of a mixed bag of Forth nuts!
>
> I also like the fact that Hugh was willing to answer questions about his code.

Yeah, when Hugh is not on a tear he can contribute something to the 
group... but it never lasts long.  Elizabeth had a stalker/attacker in 
Hawaii, but she has had one here a lot longer.  8-o

-- 

Rick

[toc] | [prev] | [next] | [standalone]

Page 1 of 2  [1] 2  Next page →

Back to top | Article view | comp.lang.forth

csiph-web