User contributions
Jump to navigation
Jump to search
- 08:27, 18 October 2007 diff hist +832 Code Racer fill out the comparison between Code Racer and Efficiency Tester per the recent hex experience
- 08:09, 18 October 2007 diff hist +20 m Talk:Hex →91+-2 milliseconds or more: add four <br/> tags to format the summary results as a table
- 08:08, 18 October 2007 diff hist +225 Hex Sorry we don't know how to format the table-of-contents correctly except by beginning with a silly empty section like this.
- 08:03, 18 October 2007 diff hist −9 m Hex strike __NOTOC__
- 08:03, 18 October 2007 diff hist +194 Efficiency Tester also link with the Code Racer harness for llGetTimestamp
- 07:57, 18 October 2007 diff hist −2 Efficiency Tester to
- 07:56, 18 October 2007 diff hist +3,230 Talk:Hex add the raw data for the scientifically reproducible summary measures of run time that now appear in the article
- 07:51, 18 October 2007 diff hist +67 m Hex fix the invisible table-of-contents -- get the = and Div tags straight
- 07:42, 18 October 2007 diff hist −98 m Hex →Measuring Concise & Conventional & Small & Fast: ouch yes let's finish that half-typed thought of how we measure fast
- 07:37, 18 October 2007 diff hist +35 m Hex reword to more clearly & briefly report the slow Efficiency Tester confirmation alongside the quick Code Racer discoveries
- 06:54, 18 October 2007 diff hist +1,208 Hex post +- 30 7 6 7 2 millisecond results of 147 105 102 113 91 at 1,000 runs of the Efficiency Tester harness for llGetTimestamp
- 05:48, 18 October 2007 diff hist +43 Efficiency Tester guess the inaccuracy is +-1ms since that's more than I saw, first time I tried measuring the cost of no code for 1000 (not 10,000) times
- 05:44, 18 October 2007 diff hist +76 m Efficiency Tester to integer max from float max, and print it as part of the result
- 05:42, 18 October 2007 diff hist +547 Efficiency Tester run at least three times always, add FIXME where the +- inaccuracy is unstated, etc., break the code into paragraphs, comment the code, apply conventional LL client four-blank indents
- 05:37, 18 October 2007 diff hist +10 m Efficiency Tester tweak comments to provide consistent whitespace, preparatory to major edit
- 05:27, 18 October 2007 diff hist +396 Hex merge the history of same & fast and different & fast implementations into the fastest posted yet, according to Code Racer
- 04:46, 18 October 2007 diff hist +168 LlGetTimestamp link with Code Racer, distinguish from Efficiency Tester and LSL Script Efficiency
- 04:31, 18 October 2007 diff hist +6 User:Ppaatt Lynagh clarify to "concise & conventional" from "clear & conventional" a la llSubStringIndex and hex
- 22:34, 17 October 2007 diff hist +52 m Hex →Implementations: further distinguish the small & fast code as clever -- reasoning by analogy with llSubStringIndex and alluding to Wikipedia on when experts optimize appropriately
- 22:33, 17 October 2007 diff hist +3 m LlSubStringIndex clarify to concise & calculates & clever from clear and tests and arcane -- courtesy reasoning by analogy with hex
- 22:24, 17 October 2007 diff hist +13 m Hex →Different & Concise & Small: clarify to "part of the conventional answer" from "part of the answer"
- 22:23, 17 October 2007 diff hist +200 m Hex →See Also: link w/ http://en.wikipedia.org/wiki/Exemplar http://en.wikipedia.org/wiki/Optimization_(computer_science)#When_to_optimize http://en.wikipedia.org/wiki/Principle_of_least_astonishment
- 22:19, 17 October 2007 diff hist −2 m Talk:Hex →Talk:Hex Archives: move outside of the accidentally enclosing section
- 22:18, 17 October 2007 diff hist +1,749 Talk:Hex →18 October Rewrite: concise/ conventional/ small may be finished, likely fast needs work
- 22:07, 17 October 2007 diff hist −108 m Hex →Alternatives: break the visually confusing parallel between English above and below the Different/ Concise/ Small section heading
- 22:04, 17 October 2007 diff hist +3 m Hex format the section headings with perfect consistency -- never any colons, always one leading and one trailing blank
- 22:03, 17 October 2007 diff hist +6 m Hex →Demo Results:: clarify to "the demo above" from "the demo"
- 22:01, 17 October 2007 diff hist +78 m Hex →Function: [[string]] hex([[integer]] {{LSL Param|value}});: correct to account for the boundary test case of the most negative integer "-0x80000000".
- 21:59, 17 October 2007 diff hist +17 m Hex to "negative or nonnegative" from "signed" since the intended meaning is "-0x" or "0x", not "-0x" or "+0x"
- 21:58, 17 October 2007 diff hist −31 Hex →Design Rationale: reword the tail of this - same meaning, less words, parallel grammar
- 21:53, 17 October 2007 diff hist +20 Hex →Different & Concise & Small: claim Concise & Small, not just Small, in this Different Alternative
- 21:48, 17 October 2007 diff hist −42 m Hex describe the llGetTimestamp benchmark with equal accuracy but less words
- 21:46, 17 October 2007 diff hist +50 m Hex →Implementations: reword to clarify, maybe
- 21:43, 17 October 2007 diff hist −1,245 Hex →Alternatives: strike the other originally uncommented alternative - guess it was an accident of conversation, of no remaining value now
- 21:42, 17 October 2007 diff hist +1,013 Hex demo the 139 bytes of bits2nybbles, taking that label from the signed hex helper function now relabelled as hexes
- 21:24, 17 October 2007 diff hist +43 m Hex →Small: add white space to paragraph the code for clarity, also add comment to explain when { value = -value } has no effect
- 21:22, 17 October 2007 diff hist +171 m Hex adopt Talk:Hex suggestion of lsn = least significant nybble, rather than lsbs
- 21:04, 17 October 2007 diff hist −796 Hex guess that the originally uncommented code since mislabeled "different & small - signed only" was actually an exemplar of smallest code
- 21:00, 17 October 2007 diff hist +1,951 Hex implement the Talk:Hex plan to reword for concise/ small/ fast for clarity, also strike the undefined term "operator" "executed"
- 19:58, 17 October 2007 diff hist +8 Hex quote the scientific repeatably measurable code size, not the astonishing plus four size naively measured by people who only add code to a nearly empty script
- 19:58, 17 October 2007 diff hist −20 m Hex ouch finish striking the undefined term LSO
- 19:48, 17 October 2007 diff hist 0 Hex →Implementations: sort the exemplars from most to least relevant to the newbie, i.e.: Clear & Conventional, then Small, then Fast
- 19:47, 17 October 2007 diff hist +229 m User:Ppaatt Lynagh thanks to Ppeennyy for http://en.wikipedia.org/wiki/Exemplar
- 19:44, 17 October 2007 diff hist +99 Hex →Alternatives: carry thru Strife's Talk:Hex suggestion to make the Demo more accessible by subordinating the less popular exemplars
- 19:38, 17 October 2007 diff hist −198 Hex →Implementations: strike the loop costs, pending a link to how we reproduce those results - see Talk:Hex
- 19:34, 17 October 2007 diff hist +651 Talk:Hex →Loop Costs Unscientific: explain strike the loop costs, pending someone making these results repeatable, thus scientific
- 19:24, 17 October 2007 diff hist +1,838 Talk:Hex →Why bother with fast or small or different code exemplars: add this explanation
- 19:21, 17 October 2007 diff hist −31,846 m Talk:Hex archive 3 Principle Of Least Astonishment, 5 Second Talk, 6 First Talk, 7 Our Growing Consensus except keep the Wikipedia Least Astonishment link
- 18:23, 17 October 2007 diff hist +272 User:Strife Onizuka/Talkpage Archive 01 ask if URU means http://en.wikipedia.org/wiki/Uru:_Ages_Beyond_Myst
- 21:48, 16 October 2007 diff hist +80 m User:Ppaatt Lynagh →Thanks: to Huney for suggesting always thinking to add sample results together with sample code