User talk:Ryan W/Sandbox/Template:Mapsortkey

Second opinion(s) requested...
The current situation is imperfect because a few pages are still categorized wrongly, and editors attempting to fix them manually get a big scary warning (multiple DEFAULTSORT calls). Obviously I'm no programmer, so I'll wait for any reactions/corrections before proceeding.

If nobody cares, my plan is to:
 * Test in Template:Transit (only impacts one article).
 * Move new template into template space.
 * Test in Template:Map and see if the site crashes. :D
 * Use new template in all other hub navboxes, even where it has no effect, so that future navboxes copypasting that markup will inherit the fixes.

Possible objections include:
 * This is unnecessary because the volume has been low enough to handle manually. (Disagree &mdash; even if the modding community vanished tomorrow, we'd have documented maybe 10% of notable levels to date.)
 * If we implement this, the previously manually fixed pages will show the red warning on ANY subsequent edit. (Yeah, my fault I'm afraid.   I'd go back and remove all those obviously.)

Thoughts anyone? Ryan W (talk) 13:11, 17 October 2015 (CDT)


 * I've tested it for various oddball pagenames.


 * E1M1 (Galaxia) -> M1 (Galaxia)
 * Hub 1: Seven Portals -> Seven Portals
 * E1M1: The Docks (Heretic) -> The Docks (Heretic)
 * MAP30: Píča (Kama Sutra) -> Píča (Kama Sutra)
 * MAP08: A Jagged Stroll (Congestion 1024) -> A Jagged Stroll (Congestion 1024)


 * Looks like the code for skipping "A" and "The" doesn't work, and it probably shouldn't cut anything for nameless levels. --Gez (talk) 13:57, 17 October 2015 (CDT)


 * I keep forgetting about that special page &mdash; thanks! The first issue is obviously fatal and I'll look at that some more.  Nameless levels could still be overridden by the name= parameter (or else add another "pos" call to check whether the distance between the colon and the open-parentheses is 1?  Yecch).  :>    Ryan W (talk) 14:33, 17 October 2015 (CDT)


 * Awfully complicated for something that COULD just be provided as an alternate template parameter when it's necessary to override the normal behavior. I have concerns from a performance POV. This will eat up enough conditions and function calls on a per-page basis to potentially pose a threat to the use of other templates on the same article - remember that limits are cumulative on a per-article basis. Not per template transclusion. --Quasar (talk) 14:34, 17 October 2015 (CDT)
 * Too bad we can't mark templates as "autosubst" or something like this. A complicated template wouldn't be too much of a problem if it was only processed once in a page's life. --Gez (talk) 14:43, 17 October 2015 (CDT)
 * That's been requested and WONTFIXed citing vandalism concerns.   Ryan W (talk) 15:12, 17 October 2015 (CDT)


 * Turns out the alternative parameter already existed, so the manual work in the current situation is less than I've asserted. (ETA: I called the other parameter "name" but it's really "noname".)  Maybe there isn't much call for this.  Maybe I at least need a slurpee before I make up my mind.  (Yes, that seems likely.)    Ryan W (talk) 15:16, 17 October 2015 (CDT)