IRC channel logs
2017-08-01.log
back to list of logs
<cmaloney>ACTION remembers all of those data statements <dsmith-work>cmaloney: Did you ever type in a BASIC program from a magazine that was just a few lines of code and about 500 data statements? (the data was actualy machine code) <cmaloney>Later Compute magazine was pretty much all "5-line-loader + 5 pages of data statements" <cmaloney>at least those had a TSR program that would prnt a checksum <cmaloney>I think Antic / Analog had those as well <cmaloney>and Byte magazine thought nothing of just printing out assembly code <cmaloney>plus they had some of the great books like "Mapping the [machine]" and "Compute's first book of [machine]" <cmaloney>I wish I had the foresight to pick up Mapping the Atari and De Re Atari back in the day <cmaloney>I have copies now, but having the sponge-like brain of a 10y/o would have been nice. <dsmith-work>One of the first things I did with the C64 was dissasemble the roms. <cmaloney>I wish I had a mentor to show me how to do that <dsmith-work>Mainly used it as a terminal to dial in to the local public-access unix system to learn C. <cmaloney>so basic was pretty much what I was used to <cmaloney>I've only read about that machine. Never saw one in person <cmaloney>Right, but you had the experience of logging into a UNIX machine and learning C in the 1980s <cmaloney>My first UNIX machine was a lab of SparcStation 1s <cmaloney>In high school we had access to a mainframe for BASIC <cmaloney>It was terrible. Stupid 3270-ish terminals that would time out based on some arbitrary load. <cmaloney>so you'd hit "enter" and it would log you out. <cmaloney>I've only played with Forth on a Jupiter Ace emulator. ;) <cmaloney>I was cogniscent enough on the 1980s to want a Jupiter Ace, but never got one <cmaloney>mostly because it looked like a neat machine <dsmith-work>Heh. My first real job was fixing 3270's. And the rest of the 360/370 stuff. <cmaloney>made the Model M keyboards seem light by comparison <cmaloney>(Keep a model M keyboard by your bed in case of intruders; you can swish it at them to scare them away) <cmaloney>At some point we moved to IBM PS/2 machines (286s... think they were model 30s) with Turbo Pascal <cmaloney>And college was all Pascal / Modula 2 / ADA (with smatterings of C, Lisp, COBOL, and assembler) <cmaloney>and I think SNOBOL, but I've looked at the Wikipedia page for SNOBOL and it doesn't look remotely familiar. <cmaloney>Hope College got a grant from the DoD to teach Modula 2 and ADA. I think ADA is a miserable language. <dsmith-work>Converted some pascal code to C but that was about it. <cmaloney>spent most of the section with ADA trying to figure out the printed manual in the lab. <cmaloney>And of course I thought Pascal was the bee's knees. <dsmith-work>Now ada seems like a language where editor templates come in really handy. <cmaloney>I also thought the 3DO was nifty, stuck with my Atari 8bit into 1993, bought a Sega Saturn and Dreamcast. <cmaloney>If I'm betting on a horse it's a fair bet that you may want to bet on something else. <cmaloney>Yeah, it would be better served with an IDE <cmaloney>Honestly I wonder how much of Modula 2 / Ada made it into Delphi <cmaloney>I think I got Delphi at a student discount at the college I worked at, but that was also when I was transitioning from Windows to Linux full time <cmaloney>where kernels were hand-compiled and disk 22 was always screwed up. <dsmith-work>cmaloney: Yeah, slackware was the first linux I installed. Kernel 1.8 (or 1.0.8?) <dsmith-work>Before I did that, needed to boot back to dos and copy packages to floppy, then boot back to linux to install them. What a pain! <manumanumanu>cmaloney: I still think pascal is a nice language. At least the deplhi Object Pascal version with all the neat things <cmaloney>manumanumanu: Right. We never got to that point. I didn't see OO in practice until after college. <manumanumanu>I think it is nicer to do object pascal development than the C family of languages <cmaloney>dsmith-work: No worries. It was a magical time for me, and I love these old machines. <dsmith-work>cmaloney: I used to dissasemble code on continuous feed paper. I've have like a 30 foot printout on the floor going into different rooms. I'm markup all the loops and functions. <ijp>the printup went uphill bothways in 20ft of snow too <cmaloney>I never got a VT-100, but I have an ADM 11 in the basement. <ijp>I keep a spectrum at home, but it's symbolic more than anything. The first "machine" I really programmed was some PIC <cmaloney>Had another terminal that literally had a radiation sticker on it. Gave that to my friend as a joke. <cmaloney><3 the ZX Spectrum, but I wasin the USa so we never had that machine proper here. <janneke>it looks like `guix environment --ad-hoc -l guix.scm' does not do what i expect? <ijp>I doubt I can help you, but you should probably clarify what you expected and what it did <mwette>Hi Folks. I am working on a project to convert .h files into .scm files in order to provide access to C libraries. In at least one case the header file is covered by GPL v2 while my code is covered by GPL v3. Does this mean those translated files cannot be released? (A specific case is for libgit2.) <civodul>mwette: IANAL, but i think that since it's an automated translation process, the liecense that applies on the source .h file also applies on the generated .scm file <civodul>it's similar when, say, a .c file is compiled to .o <mwette>OK. My thinking was to copy (by hand) the copyright notice from the header (or a representative quote from possibly multiple headers) into the ".ffi" config file. DO you think that will work? <civodul>mwette: i think you don't even need to bother <daviid>mwette: and aware of guile-cairo as well? <mwette>haha. No, I wasn't aware. I have no current interest. I was looking for header to throw at the tool (ffi-helper). But thanks. <mwette>The same issue goes for cairo, which I am interested in. I know there is a cairo-guile port but that may be a bit old. IIRC wingo produced that port. <civodul>maybe a renewed Guile-Cairo would make sense (the existing one uses the C API) <daviid>mwette: yes, guile-cairo, but it would be more productive to clone and patch... imo, and pickup another header to play with <civodul>though it's more than a bunch of .scm files, of course <mwette>I can currently convert cairo.h cairo-svg.h and cairo-pdf.h to a single .scm file which is ~ 6000 lines long; all automated. <mwette>libgit2.h converts to libgit2.scm, about 7000 lines <daviid>civodul: yeah, but guile-cairo is guile-gnome and guile-clutter 'aware', a new binding would not work ... <civodul>mwette: i guess the ffi-helper could also be used at macro-expansion time, right? <janneke>mwette: btw, thanks a lot for 0.80.4! <mwette>The idea right now is to use "guild compile-ffi path/to/foo.ffi" which generates path/to/foo.scm. <civodul>i though you could have a macro, like (make-me-an-ffi "/usr/include/cairo.h") <civodul>and that'd expand to the definitions <mwette>my script/compile-ffi.scm just calls a procedure compile-ffi-file with the spec in it. The minimal info is header name and pkg-config name. <mwette>So one could (make-me-an-ffi "cairo.h" #:pkg-config "cairo") <daviid>I wonder if cairo has a gir module <daviid>civodul: yeah, I think you are rght <civodul>mwette: anyway, impressive piece of work again! <mwette>janneke: you are welcome; IO appreciate you finding bugs. I have found more that didn't make it into 0.80.4 <janneke>davexunit: thanks, I guess my package description is broken <janneke>mwette: ah, yes rain1 found one new bug and i also just found one <janneke>making good progress with compiling tinycc <civodul>has one of you looked into compiling C to Tree-IL? <civodul>although i guess the hard part would be the run-time support <janneke>davexunit: i guess that for a guile package to work with environment --ad-hoc, it needs to set native-search-paths for GUILE_LOAD_[COMPILED_]PATH <davexunit>executables that use guile ought to be wrapped to do the right thing somehow <davexunit>civodul has brought this up before with my own software, but I don't know how to implement it <davexunit>civodul: I recall you saying how the application build system should take care of this somehow <davexunit>so that extra guix code isn't necessary, because we wouldn't want to duplicate that for every guile executable in the world <civodul>davexunit: yeah i may have said that ;-) <civodul>python-build-system does that automatically <davexunit>I couldn't tell if you mean the guix build system or the autotools build system for the application <civodul>yeah i'm not sure exactly where it'd belong <paroneayea>> SPS is a Pre-Scheme to (GNU) C compiler written in Pre-Scheme.