People say, ‘I don’t think a language like Nix should have macros’, proving they don’t know what a macro system actually means. They think it means the C preprocessor. But I just realized that Nix’s ‘inherit’ should be a macro. There is absolutely no reason it should be part of the core […]

I did the mixing of attribute sets and letrec without distinct types, but now see how to do it with an ‘assigning types to objects upon creation’ system and yet also have them recognized as types by Scheme. So I will redo it that way. When I make define-record-factory a package in its own […]

I have just been fighting with Guile’s utter inability to handle R⁷RS define-library headers correctly. And they have the gall at Guile to claim R⁷RS support is simple. That’s what really gets me. As soon as you start doing a few cond-expand in the define-library section, you learn otherwise […]

This little definition seemingly cannot be exported as part of a R⁷RS library, but otherwise works in Gambit Scheme and shows how you can use syntax-case instead of with-syntax: https://pastebin.com/DafvVsYG Because Gambit doesn’t have with-syntax. :) Just this old version of syntax-case. But […]

I wrote a ‘define-record-factory’ that does NOT directly implement any of my new ideas for type systems (they are too complicated to be worth it). But it DOES give an interface I prefer to direct use of SRFI-9. The downside is it seems impossible to implement hygienically in an R⁵RS level of […]

I just wrote this and added it to my new autoconf archive at my chemoelectric github account: https://pastebin.com/AeWDt9Yy (To be distinguished from my older ‘sortsmill’ autoconf archive, which I am phasing out.) The macro might interest people because it contains several shell commands I […]

I am so silly. How to do reversible assignment in an Icon-like Scheme sublanguage is so dang obvious. You simply wrap it in dynamic-wind.

Types are done all wrong in Scheme. It’s just wrong. You should not be testing objects for their types. You should be extracting a type object from the object and testing the type object. I am about to rework the ‘define-record-interface’ I was just so proud about into ‘define-record-factory’ […]

I got the Gerbil Scheme compiler into the Pipchix test suite but disabled by default. Because it doesn’t work in parallel testing. It doesn’t work in distcheck. And is slow. I had to completely turn off optimization, too. The developer has an arrangement for where files go during the […]

I just added (pipchix define-record-interface) and am pretty darn proud. All you need is R⁷RS/SRFI-9.

This is what they just don’t seem to get. All this nonsense with different record type SRFIs is NONSENSE. Types should be first-class objects. Things that create things should be able to assign a type to the things they create! THEN THE PROBLEM IS SOLVED GENERALLY. You want inheritance? […]

I was wrong about converting to let a=3; in let b=4; being how to do (nix-let (`a==>3)(`b==>4) in Pipchix. That would be how to do nix-let* with the asterisk! Because doing a let in order is written in Scheme with an asterisk. Without the asterisk it is a SIMULTANEOUS binding. And written […]

I watch ‘Mayday: Air Accident Investigation’ and just get more and more and more angry that our DEPLORABLE UNIVERSITIES (yes, I mean Harvard, Princeton, Oxford, etc.) graduate people who can believe FOR A SINGLE MINUTE that an airplane wing can get its lift from airflow over its TOP, or that […]

I am rearranging the record types for the AST nodes in Pipchix. And it occurs to me how WRONGHEADED it is that the record type SRFIs have ‘inheritance’ features. NO, NO, NO, NO, NO! What you do is make TYPING a thing you can handle as data just like anything else. And then you can arrange […]

I am building Gerbil and will try to add it as one of the R7RS implementations to test against. Also I may be close to figuring out one way to test with the Gambit compiler. Maybe one can compile by telling the compiler to put everything in one namespace?

How many of you folk do not know Unix had a filename length limit of 14 characters? I mean, it was a different universe. Yeah, the Berzerkeley Software Distribution embiggulated that, and then so did SysVR4.

The module system in Gambit Scheme seems to work just as I said I would make my module system work. It simply sticks the module name in the names of symbols. I have pointed out I might use a hash function to obscure the names of the modules. Just to be a scoundrel. But it is the same thing.

Alright, then. I reworked the macros to rewrite the syntax recursively if the attribute path be on the left. Thus the macros now no longer require SRFI-46/R⁷RS compliance and will work with Gambit Scheme. Gambit also seems to require that you explicitly export all macros. That is, there are no […]

The NixOS lesspipe.sh is TERRIBLE. I am done with it. I cannot wait until the end of time for a listing that then I cannot possibly read. I am writing my own lesspipe. Here it is so far. It may have bugs. https://pastebin.com/qQ82A9RQ #nixos

Yeah, that’s it. I can infer it from the Gambit manual. Gambit does not comply with SRFI-46. It must have a very old psyntax implementation as its means of implementing both syntax-case and syntax-rules. If I want to support Gambit, I will have to complicate the syntax-rules to get it to work […]

As for SRFI-148, it is easy to get that working in Gauche. But it seems fragile in other R⁷RS implementations. Being a ck-macros implementation, it should be difficult to debug. But I tried it in Chez Scheme and it complained you couldn’t even do some of the things the syntax-rules were trying […]

My syntax-rules indeed are not R⁵RS-compatible and so this MIGHT be why they do not work in Gambit Scheme, despite it claiming to support R⁷RS. SRFI-46 is needed, at least. This calls for allowing things to come after an ellipsis in a pattern. And I have that.

LOL. I was able to capture a default shell-print parameter setting for Pipchix into a continuation object (once I added Feeley-style continuation objects to Pipchix for no good reason) and later graft a procedure to that continuation. Or, in other words, save that setting for later. Which was […]

Guile seems to be the only Scheme I am testing against that does not have er-macro-transformer, either. I mean, I should expect a FEW things beyond the standards. (What may have made Gambit fail, for instance, is it may not have had certain extensions to syntax-rules that everyone else has.)

As I say, I have had bad luck with Gambit. It does not like my syntax-rules. Even though every other Scheme has no problem with them. I will not test against Gambit. I will keep it on my system, but not test Pipchix against it.

Currently I am playing with Marc Feeley’s ‘better continuations API’ using his wrappers around call/cc to implement it. Obviously not how one wants to do it. But that’s what got me to look at Gambit. Because it’s his implementation of Scheme, and also it’s an old and venerable one. Almost as […]

Hmm, it is claimed Gambit Scheme support R⁷RS Scheme now. We shall see. I have not had great luck with Gambit in the past.

There is so much confusion that I can outsmart the "geniuses" of our society on, simply by TRULY not being a dualist. Lots of "geniuses" CLAIM to have overcome dualism, without actually having done squat towards overcoming dualism. Here is something I would like people to think about, which […]

I just realized my printf encoding is also echo-compatible.

It occurs to me that for long strings you could compress them before encoding them base64. The difficulty there is writing the compressor in Scheme, so it will work with any Scheme. So it is not a small project. Nor is it an important one. I do provide, after all, a general bytevector to […]

The base64 encoder works in both R⁶ and R⁷ Google AI gave me a bunch of test cases but some with incorrect answers. :) However the ‘base64’ program gives answers. So would ‘openssl’ if one needed a second opinion.

Regarding the previous subtoot. The actual roles of the press have little to do with what the original toot I was subtooting said was the role of the press. That is a fantasy. Some press is obviously propaganda. A Marxist party’s newspaper is a propaganda sheet, however little and […]

This is a subtoot. Saying what some idealistic theory of the press says does not make that theory TRUE. If the theory be false in fact, then the theory remain false. Theories of journalism have long been loads of crap. (And don't you love my revival of the subjunctive?)

Okay I am adding (pipchix string-manipulation) And the first thing in it is base64 encoding of UTF-8 of strings. In shell scripts you can undo this with GNU coreutils. It has the ‘base64’ command. So it is one way to handle arbitrary strings, more suitable to shell scripts than JSON is. But […]

Today I solved the problem of how to represent Unicode strings in Nix. The answer should have been obvious all along. It is NOT to store them as JSON. What is the purpose of the strings? Where will they end up? IN SHELL SCRIPTS. So they should be stored AS STRINGS WITH SHELL ESCAPES. Pipchix […]

It looks as if you can add builtins to Nix, through plugins, and it looks as if it isn’t even difficult to do. So I have this devilish idea of linking Nix with Chibi Scheme. :) More practical might be link it either with TinySCHEME or with just parts of it, so Nix could read and write […]

They must not do simulator training for stalls correctly, at least not always. The correct method, I think, would be very simple. Sit the trainee down. Repeatedly set off the stick shaker and have them push the nose down. Just repeat this again and again and again. No need for scenarios […]

What really gets me about ‘Call/cc Considered Harmful’, in contrast to considering GOTO harmful, is the complaints about call/cc are calls IN FAVOR OF PREMATURE OPTIMIZATION. In other words, THEY ARE HARMFUL. If anything, that screed is an argument IN FAVOR OF CALL/CC. In other words, a […]

Okay, MIT/GNU Scheme just failed a test that everyone else passes. And it is a PITA that nobody would want to use for Pipchix, anyway. So I am dropping MIT/GNU Scheme from my test suite.

I suppose from now on I will have to obfuscate toot URLs when I copy them.

Maybe they can find new epicycles, too: https://fediscience.org/@freyablekman/115621032289772885

There, now I pass the -j option from the MAKEFLAGS to the call to ‘testsuite’. It’s just some brute force string manipulation in tests/Makefile.am. So you can get parallel tests the way you expect to get them. In Pipchix. The Makefile is mostly running tests. Well, not really. There is no […]

I see that NixOS handles CHICKEN Scheme eggs by making a lengthy CHICKEN_REPOSITORY_PATH. Therefore, as usual, NixOS takes no advantage of whatever directory caching there may be. Only this ideal of linked lists exists, for NixOS purists. That things are done differently in other OSes FOR DANG […]

My DataHand was trying to eat my code, so I turned off keyboard repeat. The same thing is a serious risk with treadmill testing. You do NOT want keyboard repeat with treatmill testing equipment. So I was assigned to turn off the keyboard repeat in MSDOS. I succeeded. Then I resigned. Then my […]

Whoa, that’s a wet snow out there.

The latest idea is that, instead of the simplified notation for quantum mechanics alongside probability theory, to convince no physicists that quantum mechanics is nothing special (because they are a cult, not a science), I will instead do this-- I will present the Aspect experiment problem in […]

Might interest: 9p - The CHICKEN Scheme wiki https://wiki.call-cc.org/eggref/5/9p #9p #scheme #plan9

I copy because I do not want to get into it. But the actual question a wise PERSON asks is whether the NAÏVE PATTERN PROCESSOR or the junior COMPUTER PROGRAMMER is the one that is capable of gaining INSIGHT. ‘THE PURPOSE OF COMPUTING IS INSIGHT, NOT NUMBERS.’ -- Hamming The ‘Computer Facts’ […]

Racket should implement Emacs Lisp, to embarrass the Guile folk who have abandoned old goals entirely. Some new Guile maintainer should come along with entirely new projects, maybe to turn it into a PL/I compiler that doesn’t do Scheme!