💾 Archived View for makeworld.gq › cgi-bin › gemini-irc captured on 2020-11-07 at 01:02:09.
⬅️ Previous capture (2020-10-31)
-=-=-=-=-=-=-
Logs from #gemini on tilde.chat The most recent 100 lines. Scroll to the bottom for the latest. ----------------------------------------- [2020-11-07T00:09:51.103Z] <boringcactus> but yeah, more on-topic-fully, the bad hacked-together Lua it took to get pandoc to spit out gemtext is probably about as bad as the bad hacked-together Python it took to get sphinx to spit out gemtext, [2020-11-07T00:09:56.809Z] <boringcactus> but. i actually know Python [2020-11-07T00:11:01.733Z] <boringcactus> both of them are built around hierarchical documents, though, and gemini is purposefully not hierarchical, so it'll always be some form of not-how-that's-supposed-to-work [2020-11-07T00:12:36.348Z] <helby> this discussion forcing me to check gopher again ;) [2020-11-07T00:15:13.204Z] <makeworld> boringcactus: You can do Python filters with pandoc using panflute [2020-11-07T00:15:20.338Z] <makeworld> https://github.com/sergiocorreia/panflute [2020-11-07T00:15:40.480Z] <boringcactus> yeah, if i had tried hard and believed in myself i could've put together more infrastructure on top of pandoc [2020-11-07T00:15:54.224Z] <boringcactus> but there's only so much you can do before it makes sense to just use a tool that's actually better for the job [2020-11-07T00:16:17.777Z] <boringcactus> like. Sphinx is built to handle highly cross-referential documentation [2020-11-07T00:41:31.150Z] * Seirdy uses md2gemini as a start, but then re-formats and edits the content a bit to flow better [2020-11-07T00:41:52.329Z] <Seirdy> some areas need to be re-worded because of the lack of linline links [2020-11-07T00:42:12.966Z] <Seirdy> s/li// [2020-11-07T00:46:20.354Z] <makeworld> Would be happy to improve md2gemini if there's a specific thing to improve :) [2020-11-07T00:46:40.357Z] <makeworld> Or at least file an issue and get to it at some point ha [2020-11-07T00:51:12.112Z] *** swinslow_ (~swinslow@c-73-89-131-136.hsd1.ma.comcast.net) quit (quit: Leaving) [2020-11-07T01:08:40.167Z] <ddevault> gemini://drewdevault.com/weekly.pdf [2020-11-07T01:08:46.886Z] <ddevault> initial pass at formatting my latest gemini article for print [2020-11-07T01:08:59.790Z] <kayw> not found :/ [2020-11-07T01:09:41.794Z] <ddevault> trya gain [2020-11-07T01:09:43.647Z] <ddevault> try again* [2020-11-07T01:09:47.188Z] <kayw> there we go [2020-11-07T01:12:21.805Z] <ddevault> somewhat improved version pushed [2020-11-07T01:13:52.156Z] <kayw> wow that looks hella nice [2020-11-07T01:15:16.945Z] <ddevault> this is the "article" template, will try my hand at something more lettors-to-the-editor-esque for gemlogs tomorrow [2020-11-07T01:24:27.250Z] *** alex11 (~alex11@bras-base-mtrlpq5302w-grc-32-76-71-172-84.dsl.bell.ca) quit (quit: Leaving) [2020-11-07T01:27:40.565Z] <raiz> just a quick question... since gemini requests should have absolute url, do you guys implement relative path translation on the client side, have a non-standard extension on the server side? [2020-11-07T01:29:06.639Z] *** alex11 (~alex11@bras-base-mtrlpq5302w-grc-32-76-71-172-84.dsl.bell.ca) joined [2020-11-07T01:29:30.220Z] <boringcactus> afaik that's usually the client's job [2020-11-07T01:29:54.661Z] <raiz> another question came to mind... [2020-11-07T01:30:56.117Z] <ericonr> raiz: before I implemented path resolution in my client I got a lot of servers telling me off for invalid urls [2020-11-07T01:31:08.818Z] <raiz> if you have a server that handles multiple domains, do you do it the HTTP way of reading the connection address or do you read the gemini request and know what domain to route to [2020-11-07T01:31:26.887Z] <raiz> what would be the proper way to do it here? [2020-11-07T01:34:02.393Z] <boringcactus> the connection address doesn't tell you anything [2020-11-07T01:34:23.711Z] <boringcactus> if your server handles multiple domains, they're probably all connecting to the same address [2020-11-07T01:34:29.795Z] <raiz> indeed [2020-11-07T01:34:30.105Z] <boringcactus> and if not, that's. an unorthodox setup [2020-11-07T01:34:41.217Z] <boringcactus> so you can lift the host from the url [2020-11-07T01:35:13.290Z] <raiz> there's way to tell what domain the accepted connection used from the server side [2020-11-07T01:35:25.457Z] <raiz> I don't know what the proper terminology here is [2020-11-07T01:35:55.180Z] * raiz reads accept(2) manpage [2020-11-07T01:36:58.920Z] <ddevault> domain or IP? [2020-11-07T01:37:09.620Z] <ddevault> domain would require a reverse DNS lookup [2020-11-07T01:37:21.661Z] <raiz> oh yeah, its in the sockaddr struct [2020-11-07T01:37:44.743Z] <raiz> no? [2020-11-07T01:37:56.081Z] <ddevault> yes [2020-11-07T01:39:11.974Z] <raiz> I'll figure it out when I start working on that part :-) [2020-11-07T01:39:36.950Z] *** nixo (~nixo@host-87-8-141-238.retail.telecomitalia.it) quit (quit: Leaving) [2020-11-07T01:42:41.298Z] <raiz> oh wait... forgot my initial question... so how do you suggest I do it, on socket layer or protocol layer? [2020-11-07T01:43:29.350Z] <ddevault> I require SNI [2020-11-07T01:43:34.569Z] <ddevault> and disconnect clients which don't offer it [2020-11-07T01:43:59.970Z] <raiz> based [2020-11-07T02:26:45.819Z] *** awalvie (~awalvie@150.242.255.20) joined [2020-11-07T02:27:14.019Z] <makeworld> ddevault: Looks not bad! I think having the arrow indented (for the xkcd link) would be good [2020-11-07T02:27:25.669Z] <makeworld> So it's not in the column margin [2020-11-07T02:44:31.382Z] *** helby (~helby@82-119-99-70.static.chello.sk) quit (quit: leaving) [2020-11-07T02:52:34.443Z] *** helby (~helby@82-119-99-70.static.chello.sk) joined [2020-11-07T02:57:08.440Z] *** awalvie (~awalvie@150.242.255.20) quit (quit: WeeChat 2.8) [2020-11-07T03:33:33.412Z] *** alex11 (~alex11@bras-base-mtrlpq5302w-grc-32-76-71-172-84.dsl.bell.ca) quit (quit: Leaving) [2020-11-07T03:38:50.412Z] *** alex11 (~alex11@bras-base-mtrlpq5302w-grc-32-76-71-172-84.dsl.bell.ca) joined [2020-11-07T04:00:00.713Z] *** avane (~avane@host86-176-156-206.range86-176.btcentralplus.com) quit (quit: ZNC - https://znc.in) [2020-11-07T04:01:31.833Z] *** avane (~avane@host86-176-156-206.range86-176.btcentralplus.com) joined [2020-11-07T04:26:25.033Z] <jcowan> What would be a good heuristic for figuring out if a file is text/gemini? I've asked on the ML too. [2020-11-07T04:27:38.060Z] <bie> jcowan: i was thinking about that the other day but couldn't really come up with anything... checking for a /^=> etc/ was the closest i got :x [2020-11-07T04:28:39.704Z] <raiz> how do they do it for markdown? [2020-11-07T04:28:45.719Z] * jcowan checks [2020-11-07T04:28:49.849Z] <raiz> if they don't then we don't, lol [2020-11-07T04:28:59.042Z] <raiz> fwiw gemtext is just plaintext [2020-11-07T04:29:12.056Z] <raiz> can be problematic for servers auto generating mime headers [2020-11-07T04:29:23.021Z] <jcowan> No support for markdown [2020-11-07T04:29:23.903Z] <raiz> but that would be worked around if necessary [2020-11-07T04:30:09.953Z] <jcowan> It can only be probabilistic, because every plain text file is valid text/gemini. [2020-11-07T04:30:51.179Z] <raiz> but not every plain text file is reasonable gemtext [2020-11-07T04:31:01.241Z] <raiz> it can be misinterpreted [2020-11-07T04:31:23.856Z] <raiz> you could use file extensions [2020-11-07T04:31:27.166Z] <bie> for my server i ended up using the extension to decide whether to serve text/plain or text/gemini [2020-11-07T04:31:29.265Z] <raiz> that's as far as it goes [2020-11-07T04:31:35.003Z] <bie> for everything else i use the mime type [2020-11-07T04:31:43.956Z] <bie> uh i mean... [2020-11-07T04:31:49.663Z] <bie> 1. libmagic checks the file [2020-11-07T04:32:08.060Z] <bie> 2. if it's text/plain check the extension, otherwise just use the mime type from libmagic [2020-11-07T04:32:27.511Z] <raiz> yeah, that's decent [2020-11-07T04:39:36.198Z] *** helby (~helby@82-119-99-70.static.chello.sk) left (#gemini) [2020-11-07T04:39:57.022Z] <bie> i was going to figure out the ipv4 problems with my gemini server today, but the nice weather is calling to me *goes outside* [2020-11-07T04:40:33.430Z] <jcowan> Alternatively, check the extension first, and use libmagic if you don't recognize it or there is none. In other words, cheapest first. [2020-11-07T04:43:00.101Z] *** rb100 (~jeff@d27-96-151-97.nap.wideopenwest.com) joined [2020-11-07T04:58:07.347Z] <khuxkm> honestly, I don't see what's so bad about just using the extension if it's there [2020-11-07T04:58:26.424Z] <khuxkm> the only place I see that being an issue is .cgi, and CGI scripts can specify their own mimetype [2020-11-07T05:10:10.567Z] <jcowan> Nothing. But not all extensions will be in the server's table, and not all files will have extensions. [2020-11-07T05:33:38.733Z] <raiz> I've noticed sending a request of "gemini://gemini.circumlunar.space" to gemini.circumlunar.space returns "31 gemini://gemini.circumlunar.space/", notice I didn't include the leading '/', is that behavior considered common? can I handle no leading slash as if it's there all along in my implementation? [2020-11-07T05:34:08.228Z] <raiz> so instead of a redirect, a request without leading slash is treated as if it had a leading slash [2020-11-07T05:34:26.505Z] <raiz> has anyone done that in their implementation? [2020-11-07T05:39:18.647Z] <khuxkm> raiz: well it's in gemini-diagnostics that you should do so (though it's more descriptive and less prescriptive [2020-11-07T05:39:21.651Z] <khuxkm> ) [2020-11-07T05:39:53.922Z] <raiz> what gemini-diagnostics? [2020-11-07T05:43:23.868Z] <raiz> actually, nvm, the redirect makes sense, there should be an explicit path [2020-11-07T05:43:28.417Z] <khuxkm> formerly jetforce-diagnostics [2020-11-07T05:43:38.179Z] <khuxkm> https://github.com/michael-lazar/gemini-diagnostics [2020-11-07T05:44:35.559Z] <raiz> sounds good [2020-11-07T05:44:45.927Z] <raiz> there's so much torture in gemini space [2020-11-07T05:48:27.226Z] *** rb100 (~jeff@d27-96-151-97.nap.wideopenwest.com) quit (quit: leaving)