Too much trouble for too little benefit. Too little benefit because it's not that urgent to link to "English" section, as the section is usually the first one, and specifying lang="en" is not needed in 99% of cases (we can use {{l}} for the rest). Too much trouble because it complicates wikicode and 'adds to mental burden of editing.' If {{l/en|...}} is an improvement, we should replace all wikilinks ([[...]]) with {{l/en|...}} in the main namespace (I don't think anyone would support this?), otherwise all of its usages should be replaced with [[...]], because it causes inconsistency. Currently we are using [[...]], {{l/en|...}} (and even {{l|en|...}}), sometimes at the same time in a single page, what a mess. --Z06:50, 12 September 2013 (UTC)[reply]
Comment. ==English== is actually not always the first section: it follows ==Translingual==. Also, for those of us using Tabbed Languages, the gadget remembers the language section we last visited, and sends us there when we visit a language-unspecified link. As for your other comments . . . for some reason, we deleted the meaningful {{onym}} in favor of the meaningless {{l}}, so I'm no longer sure. I used to think it made sense to explicitly language-tag all mentions, but since we're no longer explicitly tagging them as mentions, I guess it might not make sense anymore. —RuakhTALK06:59, 12 September 2013 (UTC)[reply]
Keep While it often isn't necessary to link to the English section, explicitly linking to it tells us that the link isn't to a "foreign"/non-English word. You don't need to tell that, but it would be nice if you allowed other to tell it. --80.114.178.722:08, 27 September 2013 (UTC)[reply]
I prefer to delete all of these l/ templates. I think the best 'low cost' way of doing it is either [[foo]] or [[foo#English|foo]]. It depends how low cost you want to go. This is somewhere in the middle where {{l}} is at the top of the range, but hopefully improved by Lua and other changes. Mglovesfun (talk) 11:16, 14 September 2013 (UTC)[reply]
Keep, as per Ungoliant. Linking to the proper section may be not so important for the longer words (which are less likely to be available in more than one language), but comes very handy for the shorter ones (as in: a.) The importance of the proper xml:lang= coding may currently be low, but I’d expect it to steadly rise as time goes by. (And why, aren’t w:Word processors of today use that information for spell-checking?) — Ivan Shmakov (d ▞ c) 14:24, 15 November 2013 (UTC)[reply]
@Acronym: The thing is, everything that {{l/en}} does can be accomplished by {{l|en}}, which Ungoliant's argument does not address. The latter template also supports |gloss= and |pos= parameters, while the former does not. Of course we could either extend {{l/en}} or switch to {{l}} every time we need them, but this increases the tedium of Wiktionary maintenance. I think I should ask a more general question here: do we still need to keep the "high-performance" language-specific templates, and then laboriously maintain feature parity (or not maintain it, adding to the cognitive load of editing)? (I failed to get a ping for some reason. Will investigate.) — Keφr16:58, 18 November 2014 (UTC)[reply]
@Keφr: I was under the impression that using the specific l/… templates was prefered to using {{l}} + the |lang= parameter at |2=; however, if that is not the case, I would be perfectly happy to switch from using a slash to using a pipe. I was going to say that we should keep {{l/grc}} because it doesn't autotransliterate, but I've recently discovered that {{l|tr=-}} suppresses autotransliteration, so I see no need to keep that one. There are some templates, like {{ja-l}} and {{ja-r}}, that AFAICT still need to be kept, but I should think that they are fairly few in number. — I.S.M.E.T.A.20:23, 20 November 2014 (UTC)[reply]
I have been under the impression for some time that the introduction of Lua has meant that the language specific "l/xx" templates are no longer necessary, because {{l}} no longer consumes as many resources as it did before Lua. I have therefore been replacing "l/xx" with "l|xx" wherever I encounter it. However, I still see other editors doing the exact opposite. Is my impression correct? Has Lua made all the "l/xx" templates unnecessary? If so, then get a bot to replace them all, and then delete them all. —Aɴɢʀ (talk) 18:35, 18 November 2014 (UTC)[reply]
When the language-specific templates were created, they were intended to be used only in the rare cases where page load was so high that we had no other option. Other people have then started to use the templates generally but I never agreed with that. I can't say whether the new version of {{l}} is faster than the old one, but it's likely that there are still a few pages where it's too slow and we will need the shortcut templates. —CodeCat19:58, 18 November 2014 (UTC)[reply]
The l/xx templates are still needed for index and frequency list pages, where there are thousands of transclusions. Lua may be far more efficient, but we make it do far more: script detection, transliteration, linking of parts in multiword terms, etc.- all impossible or impractical with templates. When you have 5,000 transclusions that have to be finished in 10 seconds, that gives you less then 2 ms per transclusion to do all that stuff. Chuck Entz (talk) 20:55, 21 November 2014 (UTC)[reply]
The current implementation of Lua is pretty inefficient though. I'm not sure but I wouldn't be surprised if every module invocation is done "from scratch", importing all the modules all over again. —CodeCat20:59, 21 November 2014 (UTC)[reply]
