[23:25:58] Krinkle: um, are you saying that every user's common.js that uses importScript() it going to break? [23:27:16] MatmaRex: see -perf. That's the one method I'm unsure about. I'm not buying the argument about there not being a replacement. All requests for a replacement are about entirely new features (e.g. wiki modules per user) which is not a replacement but a nice future for some time (something I want nonetheless but it's not a equiv-replacement). mw.loader is a [23:27:16] fine replacement. However due to the large scale use, I am willing to keep an alias around. [23:28:39] Krinkle: there is no replacement for a method that will load a page by name as script/style… i think we've discussed this the last time you attempted to kill this [23:29:05] i am not in -perf. this is not a performance matter, anyway. [23:29:09] MatmaRex: Replacement does not mean find/replace and done. We have larger deprecations all the time. This isn't what you really mean. [23:29:25] You mean the creation of an action=raw url? [23:29:39] and besides, it could be a find/replace still. [23:29:48] no, i mean loading a page by name as javascript. i don't care about the implementation details [23:30:10] What semantic magic are you referring yo? [23:30:12] to* [23:30:36] you might as well implement that functionality on top of RL and make these load.php urls, whatever [23:30:43] importScript is just importScriptURI with a 1 line convenience code to create that url. [23:31:27] Minifying arbitrary wiki pages and all that stuff are feature requests. It's not been possible before, either. [23:31:38] How is that relevant here? [23:31:43] what [23:32:01] how *is* it relevant? i don't know [23:32:09] never mentioned it [23:32:22] I assume thats' what you mean by loading it via load.php [23:32:27] nevermind that [23:32:29] gods [23:32:54] you are removing a user-facing API, where by user i mean the kind of user who edits wiki pages [23:33:26] I'm trying really hard to figure out whether you (and others, including a past version of myself) are just saying it's too much work to find/replace (which I can agree with, I said that already, I'm willling to keep aliases but trying to understand better first) - or is there actual functionality you're seeing that I don't see. [23:34:25] nobody is killing action=raw [23:35:19] You're saying we have users who write their own code (as opposed to copying instructions we can update, e.g. copy X to install X script), and those urls should not be able to construct an action=raw url? [23:35:23] is that the difference? [23:35:39] those users* [23:36:08] are you seriously saying that having "importScript('User:Matma Rex/foo.js')" is not better than "importScriptURI('https://en.wikipedia.org/w/index.php?title=User:Matma_Rex/foo.js&action=raw&ctype=text/javascript')" ? [23:36:23] or, well, mw.loader.load(…) [23:36:27] (even worse) [23:37:04] mw.loader.load('/w/index.php?title=User:Matma_Rex/foo.js&action=raw&ctype=text/javascript'); would do [23:37:14] (if you care about not hardcoding the domain) [23:37:15] it allows anyone to tell what is being loaded, to identify that script, identify its author, and remove it when an alert() pops up that this script is broken [23:37:37] buring this in a 100-character URL is mad [23:38:41] I probably care more about back-compat than most everyone. I am seriously surprised you think this is a big deal. (again, not referring to the work to update existing calls and docs). [23:39:09] and making sure that existing non-deprecated functions continue working should be the *first* thing to think about, not something to be "willing to keep" [23:39:37] alert() doesn't include source and any mention of source (e.g. script errors dialogs in Firefox and IE) will have full url either way. [23:41:59] Use case loss of functionality: None. Use case lots of work to migrate: Agreed. Use case people shouldn't have to understand: Disagreed. "People" won't write importScript() calls, they copy them them manual pages saying how to install something. Coders that write those pages seem to me like a subset of those who can craft a simple action=raw url (and already [23:42:00] have to in majority of cases due to cross-wiki compatibility, this is nothing new, many user script manual pages already refer to mw.loader.load or importScriptURI with full url for that reason). [23:42:14] The use case I hear now is something new: Debuggability for non-coders? [23:59:38] MatmaRex: If we still think the same way 4 months from now, and if we're sure that that is the main reason only, we could branch that part off wikibits into a separate file part of the base module.