Newest at the top
2025-09-09 14:11:07 +0200 | merijn | (~merijn@77.242.116.146) (Ping timeout: 248 seconds) |
2025-09-09 14:08:18 +0200 | kaotika | (~user@user/d4q) d4q |
2025-09-09 14:07:00 +0200 | kaotika | (~user@user/d4q) (Ping timeout: 245 seconds) |
2025-09-09 14:03:19 +0200 | chromoblob | (~chromoblo@user/chromob1ot1c) chromoblob\0 |
2025-09-09 14:02:25 +0200 | chromoblob | (~chromoblo@user/chromob1ot1c) (Ping timeout: 248 seconds) |
2025-09-09 14:00:53 +0200 | Lycurgus | (~juan@user/Lycurgus) Lycurgus |
2025-09-09 13:59:05 +0200 | trickard_ | (~trickard@cpe-53-98-47-163.wireline.com.au) (Read error: Connection reset by peer) |
2025-09-09 13:52:47 +0200 | Lord_of_Life | (~Lord@user/lord-of-life/x-2819915) Lord_of_Life |
2025-09-09 13:51:39 +0200 | kaotika | (~user@user/d4q) d4q |
2025-09-09 13:50:57 +0200 | kaotika | (~user@user/d4q) (Max SendQ exceeded) |
2025-09-09 13:49:38 +0200 | kaotika | (~user@user/d4q) d4q |
2025-09-09 13:47:00 +0200 | kaotika | (~user@user/d4q) (Max SendQ exceeded) |
2025-09-09 13:46:38 +0200 | <kaol> | I'll make a small example something out of this. |
2025-09-09 13:45:40 +0200 | kaotika | (~user@user/d4q) d4q |
2025-09-09 13:44:32 +0200 | ttybitnik | (~ttybitnik@user/wolper) ttybitnik |
2025-09-09 13:44:21 +0200 | <dminuoso> | Like I said, it would be easiest if you just shared your API specification |
2025-09-09 13:44:12 +0200 | <dminuoso> | But jsaddle has no custom MimeUnrender instances either. |
2025-09-09 13:44:03 +0200 | kaotika | (~user@user/d4q) (Max SendQ exceeded) |
2025-09-09 13:43:32 +0200 | <dminuoso> | fsvo "official servant package" |
2025-09-09 13:42:56 +0200 | <kaol> | I suspect there's something wrong with servant-jsaddle since servant-client-js doesn't come up with latin1. And servant-jsaddle is an official servant package. |
2025-09-09 13:41:50 +0200 | <dminuoso> | But it would outside of any of the official servant packages. |
2025-09-09 13:41:40 +0200 | <dminuoso> | Now of course maybe there's an additional type tag with a MimeUnrender instance that uses latin1 in your dependency closure. |
2025-09-09 13:41:36 +0200 | kaotika | (~user@user/d4q) d4q |
2025-09-09 13:40:03 +0200 | <dminuoso> | kaol: c.f. https://hackage-content.haskell.org/package/servant-0.20.3.0/docs/Servant-API.html#t:MimeUnrender |
2025-09-09 13:39:48 +0200 | kaotika | (~user@user/d4q) (Max SendQ exceeded) |
2025-09-09 13:39:33 +0200 | Googulator | (~Googulato@2a01-036d-0106-217b-fd1e-c506-2528-080c.pool6.digikabel.hu) |
2025-09-09 13:39:18 +0200 | Googulator | (~Googulato@2a01-036d-0106-217b-fd1e-c506-2528-080c.pool6.digikabel.hu) (Quit: Client closed) |
2025-09-09 13:39:00 +0200 | <kqr> | sm: Cheers, will do! |
2025-09-09 13:38:55 +0200 | <dminuoso> | kaol: Well whereever the mojibake happens, servant has no code path through which it even could decode latin1. |
2025-09-09 13:38:42 +0200 | kaotika | (~user@user/d4q) d4q |
2025-09-09 13:37:56 +0200 | <ski> | merijn : depends on encoding |
2025-09-09 13:37:47 +0200 | <kaol> | I found out the type already. |
2025-09-09 13:37:11 +0200 | <ski> | or `Proxy @<type>' |
2025-09-09 13:37:01 +0200 | <kaol> | I know someone's calling this decodeAs with Proxy :: Proxy JSON with latin1. |
2025-09-09 13:36:58 +0200 | <ski> | kaol : you could make it show `Proxy :: Proxy <type>' |
2025-09-09 13:36:50 +0200 | kaotika | (~user@user/d4q) (Ping timeout: 256 seconds) |
2025-09-09 13:35:51 +0200 | davidlbowman | (~dlb@user/davidlbowman) davidlbowman |
2025-09-09 13:35:34 +0200 | <dminuoso> | Im fairly sure its outside of servant. |
2025-09-09 13:35:24 +0200 | <dminuoso> | kaol: So the servant code itself does not do any latin1 decoding. |
2025-09-09 13:35:12 +0200 | <kaol> | I started with servant-client-js instead of servant-jsaddle and it didn't have encoding problems but it was incapable of handling basic authentication which made me switch. I'm starting to think that returning back and just improving its error handling is an easier path. |
2025-09-09 13:33:12 +0200 | <dminuoso> | kaol: can you share you API specification here? |
2025-09-09 13:32:10 +0200 | <kaol> | I specifically have "ö" in my test body which is in latin1 and I unpacked the body just to see the digits and it is there indeed as single byte latin1. |
2025-09-09 13:31:56 +0200 | <mauke> | it's when you send €, but you get â�¬ |
2025-09-09 13:31:16 +0200 | <kaol> | I'm using servant-jsaddle which is little used, I guess. At least the offending decode happens on servant-client-core's side. |
2025-09-09 13:31:09 +0200 | <merijn> | ah, no only for the first 127 values, which makes sense I suppose |
2025-09-09 13:30:30 +0200 | <merijn> | How can you even tell? Pretty sure latin-1 is a subset of utf-8, no? |
2025-09-09 13:30:30 +0200 | <dminuoso> | Mmm. That sounds surprising. |
2025-09-09 13:30:10 +0200 | <kaol> | Well it has a bug. Content-Type is correct. |
2025-09-09 13:30:10 +0200 | <dminuoso> | The actual encoding is irrelevant, there's no heuristics. |
2025-09-09 13:29:57 +0200 | <dminuoso> | kaol: It respects Content-Type. |