What kind of decoupling did you have in mind?
Not specifying that each skin has to have exactly one lc identifier and then starting to rely on this requirement and generate all sorts of secondary names, identifiers, paths, class names, etc. from that. E.g why not just ask that skin for it's localized name?
I second this, code (skin or extension) should be expressive and if possible be decoupled. Doing all sorts of magic behind a curtain may save some line of code but it certainly does not improve readability or expressiveness and makes it prone to breakage if some of the "magic" disappears.
On 6/2/14, Stephan Gambke s7eph4n@gmail.com wrote:
On 1 June 2014 22:45, Daniel Friesen daniel@nadir-seen-fire.com wrote:
What kind of decoupling did you have in mind?
Not specifying that each skin has to have exactly one lc identifier and then starting to rely on this requirement and generate all sorts of secondary names, identifiers, paths, class names, etc. from that. E.g why not just ask that skin for it's localized name?
I know there is loads of legacy code to deal with here and this business with the message identifiers for the skin names in particular is not the object of the on-going changes. It's just that I'd rather not have an explicit requirement introduced specifying that there must be exactly one all-purpose lower-case id per skin.
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l