Anonymous user
Talk:Josephus problem: Difference between revisions
m
→Categorisation of solutions: Indices.
ReeceGoding (talk | contribs) No edit summary |
m (→Categorisation of solutions: Indices.) |
||
| (11 intermediate revisions by 4 users not shown) | |||
Line 45:
:::::: I am asking for programs that run unmodified on all versions and that is quite easy. REXX->PL/I and others I'd call translation! REXX->ooRexx is also a translation when I use new features in order to simplify the algorithm. --[[User:Walterpachl|Walterpachl]] ([[User talk:Walterpachl|talk]]) 08:44, 11 May 2013 (UTC)
:::::
:::::::: All that is needed to see if a REXX program works under a Classic REXX is to (simply) run/execute the REXX program under a Classic REXX interpreter. That would solve your (above) problem of ''it '''should''' work on all REXXes''. Just executing a REXX program under ooRexx doesn't mean that it works under Classic REXX as well. This seems like an obvious and simple thing to do. If you can't/won't use a Classic REXX interpreter and you just have ooRexx, then enter that program under the ooRexx section. That will ensure that your program will run/execute correctly (verifiably) under the appropriate language section. -- [[User:Gerard Schildberger|Gerard Schildberger]] ([[User talk:Gerard Schildberger|talk]]) 18:54, 27 June 2020 (UTC)
Many of GS's REXX programs won't work on ooRexx because they use the things I mentioned: $ etc. as variable names and x= instead of x="" for assignment of the null string. When I try one of these programs I make the necessary little changes (c/$/d/* * and add the "") and I am off (and learn a lot from them)
Line 52 ⟶ 54:
I was going to add a considered comment but I can't be bothered in reopening the Rexx/ooRexx religious war; life's too short. I choose to ignore this trite argument about what is and what isn't Rexx --[[User:Alansam|Alansam]] ([[User talk:Alansam|talk]]) 00:32, 10 May 2013 (UTC)
: In calling these arguments trite is just belittling the differences and ignore the substance of the issue(s). The issue (argument) about what is and what isn't Rexx is a strawman argument. The issue is: what is and isn't '''<u>Classic</u> REXX'''. -- [[User:Gerard Schildberger|Gerard Schildberger]] ([[User talk:Gerard Schildberger|talk]]) 18:54, 27 June 2020 (UTC)
==Formulae hidden to most browsers by under-tested cosmetic edits at 17:36, 21 April 2016 ==
Line 122 ⟶ 125:
::Thanks! That definitely sounds right. --[[User:ReeceGoding|ReeceGoding]] ([[User talk:ReeceGoding|talk]]) 16:13, 24 June 2020 (UTC)
::I think that you might also have explained what the jq solution is doing. --[[User:ReeceGoding|ReeceGoding]] ([[User talk:ReeceGoding|talk]]) 16:41, 24 June 2020 (UTC)
This certainly piqued my interest! Factor notes above will help me too. I have spent the day scouring the entries and am left with Haskell, Python[4 aka learning iter in python], REXX[version 2] as the last three I cannot get (plus Befunge, J, and Mathematica which I am quite happy to ignore). This is my analysis so far:
;1. skipping<nowiki>:</nowiki> 360 assembly, 6502 Assembly, AWK, EchoLisp, ERRE, MATLAB, NetRexx, Phix, PHP, PL/I, REXX[version 1].
:Method: all prisoners stay where they are, executioner walks round and round, skipping over more & more dead bodies (slowest of the lot, by quite some margin)
;2. linked list<nowiki>:</nowiki> AArch64 Assembly, Ada, ARM Assembly, Common Lisp[2, probably], Fortran, JavaScript[1] (albeit dbl-lnk), Python[3].
:Method: like skipping, all prisoners stay where they are, but the executioner uses the links to speed things up a bit.
;3. sliding queue<nowiki>:</nowiki> Clojure, Crystal, D (both), Eiffel, Elixir, Erlang, friendly interactive shell, Go, jq, Perl, PowerShell, PureBasic (albeit one at a time), Raku, REBOL, Ruby, Scala, Sidef[1], Tcl.
:Method: all skipped prisoners rejoin the end of the queue which sidles left, executioner stays put until the queue gets too short.
;4. contractacycle<nowiki>:</nowiki> AppleScript[2], Groovy
:Method: executioner walks along killing every k'th prisoner; queue contracts to remove gaps while he walks back. (reverts to one at a time once queue too small, as next)
;5. contractalot<nowiki>:</nowiki> AutoHotkey, C#, C++, Frink, Formulae, Java (both), JavaScript[2], Julia[2], Kotlin, Lua, NanoQuery, Nim, Objeck, Oforth, Processing, Python[1], R[2], Rust, Seed7, Swift, VBScript, Vedit, VisualBasic.NET, XPL0, zkl.
:Method: executioner walks round and round, queue contracts after every kill. (This is a mix of your 3 and 5, but not MATLAB or REXX or AWK)
;6. recursive<nowiki>:</nowiki> Emacs Lisp, Icon, Julia[1], PARI/GP, PicoLisp (less the optms.n), Sidef[2] (Your 4, but not Erlang)
:Method: recursive mod maths madness - only handles the lone survivor case.
;7. iterative<nowiki>:</nowiki> ALGOL 68, ANSI Standard BASIC, AppleScript[1,3(!!)], BASIC, Batch File, C (but not ULL), Common Lisp[1], Factor, Forth, FreeBASIC, Modula-2, Python[2], R, Racket, Ring, SequenceL, ZX Spectrum Basic
:Method: iterative mod maths madness - but hey, it will be extremely fast. Unlike recursive, it can also deliver >1 survivor, one at a time. (Your 6)
;8. iterative2<nowiki>:</nowiki> Icon[2]
:Method: What do you call a deer with no eyes? ... no idea
I will be quite shocked if there are no glaring errors in the above. I was gobsmacked to realise that AppleScript[3] (aka composition of pure functions) is actually the exact same algorithm as AppleScript[1]!!! (well, just the survivor part is, I should say) Have now posted all 8 algorithms in Phix, just to annoy everyone (they are all quite short). --[[User:Petelomax|Pete Lomax]] ([[User talk:Petelomax|talk]]) 03:21, 27 June 2020 (UTC)
: The Forth solution is another iterative mod maths solution like R, except it uses hard coded input. (Less stack shuffling that way?) --[[User:Chunes|Chunes]] ([[User talk:Chunes|talk]]) 04:07, 27 June 2020 (UTC)
:Very impressive! Studying this will keep me busy for days.--[[User:ReeceGoding|ReeceGoding]] ([[User talk:ReeceGoding|talk]]) 18:16, 27 June 2020 (UTC)
::In the new R solution, I found it odd that you wrote
It is 1-indexed, meaning that we will have a tough time using most solutions that exploit modular arithmetic.
::(1) it is just below the existing R "Growing circle solution" which uses modular arithmetic, and
::(2) Phix has 1-based indexes and it didn't bother me. Anyway, I added R[2] to "contractalot". --[[User:Petelomax|Pete Lomax]] ([[User talk:Petelomax|talk]]) 20:43, 30 June 2020 (UTC)
::The previous solution is part of the reason why I said "most". As for 1-indexing causing problems, it's mostly to do with how you constantly have to account for modular arithmetic wanting to make things that your language wants to be 1 in to 0. It can be worked around, but it's yet another difficulty in an already hard problem.--[[User:ReeceGoding|ReeceGoding]] ([[User talk:ReeceGoding|talk]]) 21:35, 30 June 2020 (UTC)
:::(Modular arithmetic. Excuse me as I nab this for another reason supporting languages needing to start list indices from 0 😊) --[[User:Paddy3118|Paddy3118]] ([[User talk:Paddy3118|talk]]) 23:16, 30 June 2020 (UTC)
| |||