Pocket Lisp Machine
in [Lisp]
|
From: GP lisper on 25 Sep 2006 20:52 On 25 Sep 2006 16:00:03 -0700, <nallen05(a)gmail.com> wrote: > GP lisper wrote: > > but there are also much more mature c->6800 and c++->6800 compilers. > Would it be easier to use figure out how to interact with the OS with > thinlisp or ecl and compile the generated c to 6800 with a tried and > true c->6800 compiler? I'd have to learn C, something I've successfully ducked for over a decade, while Assembly is an old friend. In any case, I'd want some source that had a small core in assembler, with everything else built up from lisp. I can grab the lisp source pieces out of Codemist for instance, and drop them on top of an assembler base. > this all sounds very shady ;-) Hey, that reminds me that maybe we should be announcing those Reno Lisp Users meetings in this newsgroup, as some of the other westerners (even the cold ones) have been doing. The only thing that it's sounding to me is that of a long winters project. -- Reply-To email is ignored. -- Posted via a free Usenet account from http://www.teranews.com
From: D Herring on 26 Sep 2006 00:21 GP lisper wrote: > I have an application that needs lisp and some stealth[1]. Recently > I've been looking at TI calculators, since they have some interesting > specs, in this case a 68000 processor, the same as in early Macs. At least get a decent machine: http://www.shopping.hp.com/webapp/shopping/product_detail.do;jsessionid=FYaRX9nWabh2wBI1PycEjHtXhkLK3IE1n43HDU9JZVeZUU1oYKbG!1113141265?storeName=storefronts&landing=handhelds&category=calculators&orderflow=1&product_code=F2229AA%23ABA&catLevel=1 70MHz ARM7 CPU, 2.5 MB RAM, SD card slot, ... all for $150. And they appear to have thrown in a decent keyboard to boot. See comp.sys.hp48, http://www.hpcalc.org/, and http://hpgcc.org/
From: Ari Johnson on 26 Sep 2006 00:44 GP lisper <spambait(a)CloudDancer.com> writes: > On 25 Sep 2006 16:00:03 -0700, <nallen05(a)gmail.com> wrote: >> GP lisper wrote: >> >> but there are also much more mature c->6800 and c++->6800 compilers. >> Would it be easier to use figure out how to interact with the OS with >> thinlisp or ecl and compile the generated c to 6800 with a tried and >> true c->6800 compiler? > > I'd have to learn C, something I've successfully ducked for over a > decade, while Assembly is an old friend. In any case, I'd want some > source that had a small core in assembler, with everything else built > up from lisp. I can grab the lisp source pieces out of Codemist for > instance, and drop them on top of an assembler base. Don't be put off. C combines the power of assembly with the ease of use of assembly. :P I think the easiest thing to do here is build a runtime, gc, and image-loader in assembly and then define the Lisp you want to use. Once you have that Lisp defined, write the compiler for it in the same dialect, preferably a subset of Common Lisp or Scheme to ease cross-compilation in the bootstrapping process. Then write whatever functions you need in Lisp or assembly, depending on which is easier for each function. Pack them into an initial image and dump it to the calculator. That was my plan, at least. And it would have worked perfectly if it hadn't been for those darn meddling kids. I got as far as the runtime and the beginning of the GC, including a calling convention. How far I got depends whether you ask me about the TI-85 or TI-89. I wrote some code for each toward the same goal. I'd be interested in helping out in such a project, if anyone undertook it. I do think that rolling your own Lisp is the quickest way to get what you want on the calculator.
From: GP lisper on 26 Sep 2006 05:44 On Tue, 26 Sep 2006 00:44:05 -0400, <iamtheari(a)gmail.com> wrote: > GP lisper <spambait(a)CloudDancer.com> writes: >> >> I'd have to learn C, something I've successfully ducked for over a >> decade, while Assembly is an old friend. In any case, I'd want some >> source that had a small core in assembler, with everything else built >> up from lisp. I can grab the lisp source pieces out of Codemist for >> instance, and drop them on top of an assembler base. > > Don't be put off. C combines the power of assembly with the ease of > use of assembly. :P A->B->C, and that A stood for assembler in the true history of unix, i.e. prior to "designed by computer scientists worrying about their place in history". All the bit pushers I knew kept saying that their macro libs were better than early Cs, and muttered about the slow and bloated code they produced. > I think the easiest thing to do here is build a runtime, gc, and > image-loader in assembly and then define the Lisp you want to use. > Once you have that Lisp defined, write the compiler for it in the same > dialect, preferably a subset of Common Lisp or Scheme to ease > cross-compilation in the bootstrapping process. Then write whatever > functions you need in Lisp or assembly, depending on which is easier > for each function. Pack them into an initial image and dump it to the > calculator. Nooooo, that would create a 'development' environment on this limited hardware and ignore that super-duper computer nearby. My purpose just needs a 'runtime' environment, any code can be created on a nearly dual-xeon system. A very calculator like problem is involved for a limited deployment environment, however the code that runs daily will probably not be the same as yesterdays code, so a smart lisp base in the calculator seems perfect. So the 'compiler' for it is totally external and can be common lisp, with an output into something like clisp makes perhaps. As long as the 'calculator program' primatives are supported on the Pocket Lisp Machine, it should be fine and super-speed is not necessary, an answer that takes a couple of minutes seems OK. So you are right, I need the image-loader and a bootstrap permanently in the calculator plus a basic BIOS, but the GC and runtime exist inside that externally produced image. This is just like CP/M, you loaded the code at 100h, jumped to it and the BDOS prayed it would 'return'. > I'd be interested in helping out in such a project, if anyone > undertook it. I do think that rolling your own Lisp is the quickest > way to get what you want on the calculator. I agree that 'rolling your own' is the true path here and can see that the problems seem doable within the above approach. But a 160x100 dot-matrix screen seems deathly small. That's about 5 words per line, and 14 lines and 70 words per page. I'm used to 32 words per line. -- "Because no one uses Lisp, I'll have my pick of thousands of great, experienced programmers looking to work for someone with a non-zero IQ" --Richard Gabriel -- Posted via a free Usenet account from http://www.teranews.com
From: GP lisper on 26 Sep 2006 05:56
On Tue, 26 Sep 2006 00:44:05 -0400, <iamtheari(a)gmail.com> wrote: > > That was my plan, at least. And it would have worked perfectly if it > hadn't been for those darn meddling kids. I got as far as the runtime > and the beginning of the GC, including a calling convention. How far > I got depends whether you ask me about the TI-85 or TI-89. I wrote > some code for each toward the same goal. I forgot to mention that you should see the Codemist source at http://homepage.ntlworld.com/codemist/csl/index.html Look in ccl/src/cslbase/0notes -- Reply-To email is ignored. -- Posted via a free Usenet account from http://www.teranews.com |