[18:41:11] Доброго дня [21:05:44] I may have found a possible API bug [21:05:46] action=parse&format=php&text=%7B%7Bgoogle+books+%7Cplainurl%3D+yes+%7Cid%3D+mIT5-BN_L0oC+%7Cp%3D+137+%7D%7D&contentmodel=wikitext [21:06:04] fails to acknowledge that there's an external link present [21:06:39] so https://en.wikipedia.org/w/api.php?action=parse&format=php&text=%7B%7Bgoogle+books+%7Cplainurl%3D+yes+%7Cid%3D+mIT5-BN_L0oC+%7Cp%3D+137+%7D%7D&contentmodel=wikitext [21:07:23] https://en.wikipedia.org/w/api.php?action=parse&format=jsonfm&text=%7B%7Bgoogle+books+%7Cplainurl%3D+yes+%7Cid%3D+mIT5-BN_L0oC+%7Cp%3D+137+%7D%7D&contentmodel=wikitext is formatted better [21:09:05] Also for some reason it creates an arbitrary page and pageid [21:09:44] Anyone? [21:16:12] Anyone? [21:17:58] Cyberpower678: that template doesn't make a real link does it? It looks like it just makes a url [21:18:18] bd808, what's the difference./ [21:18:38] between a URL and an anchor tag? [21:18:44] bd808, ? [21:19:02] [21:19:05] The MW software recognizes bare URLs as external links [21:19:36] not in that context though. You'd need to use that output in another render [21:19:43] ? [21:19:50] like the example at https://en.wikipedia.org/wiki/Template:Google_books [21:20:18] If I were to paste that URL as is in my sandbox it would become a link in the rendered output. [21:20:55] I don't think it will becuase it it protocol relative [21:21:05] https://en.wikipedia.org/w/api.php?action=parse&format=jsonfm&text=//foo.bar&contentmodel=wikitext [21:21:21] but with a protocol -- https://en.wikipedia.org/w/api.php?action=parse&format=jsonfm&text=http://foo.bar&contentmodel=wikitext [21:21:33] Erm. Should MW be able to pick those up, even with a missing http? [21:21:43] *Shouldn't [21:21:50] so any arbitraty strign starting with // is a url? [21:22:34] If it parses as a valid. [21:23:21] protocol relative urls are a browser hack. they aren't really urls except in context. A real URL is scheme://path [21:23:56] without "scheme:" it's only recognized as an URL in context in a DOM [21:24:14] bd808, they're not really a broswer hack anymore. it gives the user control which protocol to use. [21:24:38] it gives the embedding page control, not the user [21:25:36] bd808, if the user prefers to browse http, then they benefit the embeded URLs missing the scheme. [21:26:05] "Protocol-relative URLs only make sense in hyperlinks in webpages. Once printed on paper, they are no longer protocol-relative." -- https://en.wikipedia.org/wiki/Wikipedia:Protocol-relative_URL [21:26:28] Meh [21:26:39] but it becomes a context sensitive parsing problem [21:27:06] you need to be expecting an URL to recognize that you have a string that is sort of a URL if you add the current scheme to it [21:30:53] * Cyberpower678 implements a hacky approach. [21:30:58] Sounds like you're talking about https://phabricator.wikimedia.org/T54253 [21:32:26] Nemo_bis, That thing is 3 years old. [21:37:40] so? [21:37:44] https://twitter.com/OSM_Tech/status/713176035831971840 [21:56:28] [[Tech]]; 105.133.176.179; [none]; https://meta.wikimedia.org/w/index.php?diff=15477236&oldid=15470197&rcid=7575038 [22:10:26] [[Tech]]; Ah3kal; Reverted changes by [[Special:Contributions/105.133.176.179|105.133.176.179]] ([[User talk:105.133.176.179|talk]]) to last version by Nemo bis; https://meta.wikimedia.org/w/index.php?diff=15477239&oldid=15477236&rcid=7575041