[11:27:09] Hi! How should Oauth keys be saved securely on Toolforge? [11:33:08] If you're saving them in a file, ensure that your tool owns the file, and that only the owner has the right to read the file (e.g. via chmod set the mode to be 400) [11:33:24] And that the file is not in a directory that served by the webserver [11:35:10] Smart! Thanks [11:35:54] That's kind of the generic advice for unix based hosts. I assume it all applies to toolforge, but i have no idea if toolforge has anything specific to itself [12:23:44] how is this possible (2 edits in bottom were made at 2:23 and 2:24 in local time before the time zone changed)? https://phabricator.wikimedia.org/F31711052 [12:26:39] Is it just me, or does the output of graphoid look blurry, particularly the font rendering? [12:26:59] Almost like the png was resized without applying sharpening [12:27:23] I think it'd look much better with some post-processing to sharpen. Or maybe just using high-dpi images [12:33:52] Maybe something is wrong with the font-hinting... [14:17:46] yeah, there's definitely something up with the rendering [14:24:26] I wonder if it's related to the existing PNG thumbnailing issues https://phabricator.wikimedia.org/T192744 [14:25:48] AntiComposite: Those are mostly caused by using a different process for jpeg and png images though [14:25:58] Which i don't think applies to graphoid [14:26:06] but honestly, I have no idea what graphoid actually does [14:26:56] There is a generally rule that resizing algorithms makes images look less sharp. Usually sharpening photos after resize makes it look better, but the same usually isn't true for graphs and such [14:27:58] https://www.mediawiki.org/wiki/Extension:Graph#Known_bugs_&_limitations [14:28:07] oh nice, it's been a known bug and limitation for a while [14:28:32] (if that bluriness is this bluriness) [14:28:36] hmm [14:28:55] Maybe related, but it looks blurry to me even when viewing the image directly [14:29:06] which wouldn't have browser upscaling [14:35:53] As an aside, the graph extension really does not degrade gracefully, if you have anti-canvas fingerprint resistence stuff enabled... [15:00:07] https://phabricator.wikimedia.org/T156191 [15:42:30] AntiComposite: I just wrote an essay about how i feel about the graphs extension https://www.mediawiki.org/wiki/User:Bawolff/Reflections_on_graphs [15:45:37] yeah, that all makes sense to me [15:50:25] the problem I see with the graph extension is that it takes a lot of work to set up, and then doesn't produce a result as nice as what I could do in a graphics editor [15:51:35] Using https://en.wikipedia.org/wiki/Template:Graph:Map for more choropleth maps might be interesting, and would make it easier for other editors (who aren't comfortable hand editing an SVG) to update them [15:51:55] * AntiComposite wonders if there's administrative level 2 map data available already [15:53:00] yeah. If its easier to just upload a file, why wouldn't people do that [15:53:31] it would be nice if that template worked more like the SVG CSS styling: instead of |US=red, |GB=red, |FR=blue; have |red=US, GB, |blue=FR [15:53:44] It also seems really hard to get the results just right. Often labels slightly overlap, or large numbers don't have thousands separators, etc [15:54:46] The kartographer maps would actually be a better idea in a lot of cases, but editors seem to hate them for some reason [15:55:05] * AntiComposite notes that those editors are not the ones maintaining the existing maps [15:56:48] hmm, why do they hate kartographer? [15:57:06] too interactive, and it says openstreetmap on it [15:57:21] How dare we give credit where credit is due! [15:57:42] Complying with the explicit requirements of a license, what horror! [15:59:00] there's also a case, which is probably true, that most maps don't need a streetmap baselayer. [16:00:29] Hmm, April fools soon, maybe as an april fools joke, we should add support for https://timqian.com/chart.xkcd/ [16:01:29] that looks really bad compared to the matplotlib implementation [16:01:51] which, as far as I'm concerned, completely nailed the style [16:02:42] Hmm, you're right that does look much better [16:03:09] Hmm, does https://phabricator.wikimedia.org/T248792 imply that WMF is un-abandoning the graph extension? [16:04:51] meh, who knows. Probably not [16:06:00] https://phabricator.wikimedia.org/T242855#6006349 [16:07:30] AntiComposite: Well, https://phabricator.wikimedia.org/T211881#6006347 might override that [16:08:28] Not really clear if milimetric is adopting it as part of his job or in a volunteer fashion because he's nice and likes graphs (implication seems to be more the latter) [16:08:51] mmm. I wonder if they ever got the contractor mentioned in https://phabricator.wikimedia.org/T236892 [16:14:06] On the other hand, undeploying graphoid might make the graph extension better, since the font-rendering is so bad in the graph extension [16:14:19] when using graphoid [16:14:29] aye [16:14:55] the existing graphoid appears to have svg support as well, but we don't use it [16:48:01] bawolff / AntiComposite: I’m going to maintain graphoid officially with my team’s approval for the short term and then once it’s clean and shiny try to find an owner. If I don’t I’ll watch over it as a volunteer. Thanks much for the thoughts. [18:31:42] [[Tech]]; 77.121.152.181; /* How system understand that 03/29/20 is 29 March? */ new section; https://meta.wikimedia.org/w/index.php?diff=19936604&oldid=19930711&rcid=15151092 [18:31:47] [[Tech]]; Hamish; Undid edits by [[Special:Contribs/77.121.152.181|77.121.152.181]] ([[User talk:77.121.152.181|talk]]) to last version by Matěj Suchánek; https://meta.wikimedia.org/w/index.php?diff=19936605&oldid=19936604&rcid=15151094 [18:31:51] [[Tech]]; Hamish; Undid edits by [[Special:Contribs/Hamish|Hamish]] ([[User talk:Hamish|talk]]) to last version by 77.121.152.181; https://meta.wikimedia.org/w/index.php?diff=19936606&oldid=19936605&rcid=15151096 [18:33:01] [[Tech]]; Dimon2712; /* How system understand that 03/29/20 is 29 March? */; https://meta.wikimedia.org/w/index.php?diff=19936608&oldid=19936606&rcid=15151170 [21:52:53] Wait, is MW installer broken in git master, or is it just me [21:57:36] oh, git pull fixed it