#linuxcnc-devel Logs

Feb 16 2018

#linuxcnc-devel Calendar

10:49 AM seb_kuzminsky: dxf2gcode is now in sid: https://packages.debian.org/sid/dxf2gcode
12:08 PM jepler: great news! and big thanks to the developers and packagers, if they're listening
01:23 PM skunkworks: sed?
01:23 PM skunkworks: sid
01:23 PM skunkworks: I thought the next version was buster?
01:24 PM skunkworks: oh - sid is always the devel version?
01:25 PM seb_kuzminsky: yeah
01:25 PM seb_kuzminsky: unstable, "sid's the kid that breaks your toys"
01:25 PM skunkworks: duh
01:30 PM seb_kuzminsky: huh, in related news it looks like anders wallin has started working on opencamlib again
01:57 PM skunkworks: look squirrel!
02:02 PM seb_kuzminsky: heh
04:39 PM jepler: hmmm the sequence Ad/n gives 1, but at the start of the program a shorter sequence is "Az" or ".z". My program doesn't check "z", which determines the depth of the stack; I didn't realize it was useful. A total of 16 best programs have the substring Ad/n, all at the start of the program.
04:40 PM jepler: and "zn" is a shorter program to print 0 but that only helps with zero itself
04:47 PM skunkworks: jepler, what are you doing?
04:47 PM jepler: skunkworks: nerding out
04:48 PM jepler: skunkworks: the program 'dc' is a RPN calculator that has been standard on UNIX for a long time -- esolangs.org/wiki/Dc says "originally written in assembler for Unix V1"
04:48 PM jepler: skunkworks: someone once (in late 2013 / early 2014) proposed the question: what's the shortest program in dc, *WHICH CONTAINS NO DIGITS 0-9* and prints "2014"
04:49 PM skunkworks: ah - cool!
04:49 PM jepler: skunkworks: for some reason that question has been rattling around inside my brain for years, and so I've tried to write a program that, for each number 0..999999, will find a short program for dc that prints that number
04:49 PM skunkworks: I have used database programs to solve puzzles... ;)
04:49 PM jepler: for example, "A" stands for the number ten, and "n" prints the most recent value, so "An" is a cood candidate for the shortest program to print "10"
04:52 PM jepler: so anyway over the last few weeks I've written a series of programs to try lots of programs and record the shortest one found that gives each number .. but since there are a lot of possible programs, one of the things I did was make guesses about which letters would even figure in a "shortest program" for a particular number. For instance, the letter "c", to discard all the values in the RPN
04:52 PM jepler: stack, probably doesn't...
04:53 PM jepler: so I was just lamenting the discovery that "z" (which pushes onto the stack the number of things on the stack) *DOES* give a shorter program for some numbers, but my search program doesn't ever try dc programs that contain "z".
04:54 PM skunkworks: awesome :)
04:55 PM jepler: my latest search program has tested about 15.3 billion dc programs, and has discovered programs without digits that print 562018 of the numbers 0..1000000.
04:56 PM jepler: it's currently testing 10-character programs that begin with 'C'
04:56 PM jepler: anyway, it keeps me off the streets, but it's pretty esoteric
05:23 PM skunkworks: some people ice fish...
05:38 PM jepler: ooh and here are some values shortened by "a"
05:39 PM jepler: ugh don't want to add "a" to my mini dc implementation
05:40 PM jepler: (e.g., "CCEdnan" and "AFd*BB+n" both give "13446" as output)