Anonymous user
User talk:NevilleDNZ: Difference between revisions
m
→Use of works with template: regarding code bit rot.
(→Use of works with template: new section) |
m (→Use of works with template: regarding code bit rot.) |
||
Line 36:
Re [http://rosettacode.org/mw/index.php?title=Roots_of_unity&diff=81425&oldid=prev]: The works with template should be used ''only'' if the example uses some sort of "nonstandard" features or runs into common implementation bugs. If the example works with any correct implementation of the language specification, and correct implementations are not uncommon, then works with should be omitted. —[[User:Kevin Reid|Kevin Reid]] 13:41, 2 May 2010 (UTC)
It seems to me that if (when) the 'standard' implementation of a language evolves, then various code specimens will spontaneously break. This kind of "bit rot" can be avoided by providing a simple test case, and a hint to which was the "works with"/"last known good" implementation/release. This is particularly important with languages that are evolving. But even well defined and static languages there are areas that have been missed by the standard definition (IF they even have a standard definition). These issues will only be discovered if the somewhere there is reference to which implementation/release was used when the code specimen was submitted.
Also as the codes is run on another CPU, then the code specimens can break, or output results may actually be required to vary. Again a hint as to the implementation/CPU should be hinted out somewhere.
Previously I have had numerous problems with python code samples that use the newer language features. C code often generates different results on 8 bit CPUs with 16 bit ''ints''. I have encountered a couple of example with Algol68.
It reminds me a bit of the "[http://www.navy.mil/navydata/questions/litehuse.html lighthouse urban myth]". Your call!
[[User:NevilleDNZ|NevilleDNZ]] 14:21, 2 May 2010 (UTC)
| |||