On 20-Jan-2015, Don Armstrong wrote: > Those files which are in the scowl binary package are [in encodings > that are not UTF-8], but that's because that's how upstream > distributes them. > > It makes sense for them to also be in UTF-8, but I'm loathe to > change them without fully understanding the use cases of people who > have scowl installed, and whose usage might be broken by a sudden > switch from ISO8859-1 to UTF-8.
Hmm. Could you transition the package, by providing in the next release: * /usr/share/dict/scowl/ No data files, only the directories below. * /usr/share/dict/scowl/ISO-8859-1/ The upstream data files in the upstream's choice, ISO-8859-1 encoding. (If additional non-Unicode encodings are provided, additional directories can provide those too.) * /usr/share/dict/scowl/UTF-8/ The same data files, transcoded to UTF-8. This would signal the change, while allowing allow any dependent programs to make an explicit choice about which encoding they want. Then, in some future release, the UTF-8 files can be the default, while the non-Unicode encodings can continue to exist. Then, and only if this seems desirable, we can deprecate the non-Unicode encodings; and finally, in some release after that, remove any non-Unicode encodings. Does that sound feasible? -- \ “A computer once beat me at chess, but it was no match for me | `\ at kick boxing.” —Emo Philips | _o__) | Ben Finney <b...@benfinney.id.au>
signature.asc
Description: Digital signature
