| 12 Sep 2025 |
 magic_rb | Aaaah DNS is cursef | 14:03:19 |
 K900 | Technically not really an issue anymore in many ways because browsers ship the public suffix list | 14:03:41 |
| K900 | And can consult it and avoid this kind of nonsense | 14:03:47 |
| K900 | But it was a concern at the time of writing the RFC and it was never updated since | 14:04:01 |
 Jassuko | Public Suffix List is problematic as well. :D | 14:04:39 |
| K900 | In other ways, yes | 14:05:02 |
| Jassuko | Yeah. It would not need to exist, if the cookie spec was sane.. I think it's just rabbit holes all the way down at this point, lol. :D | 14:05:45 |
 toonn | K900: That applies to sub.host but not to sub.host.local, not appending ".local." would still "just work." | 14:06:11 |
| K900 | Yes, except you could not know when not to append .local | 14:06:54 |
| K900 | Before the PSL existed | 14:06:54 |
| toonn | I don't see how subdomains make that more problematic though. Just never append if the hostname has a ".?" | 14:08:39 |
| K900 | I mean, you're welcome to propose a spec change if you want this so much | 14:13:43 |
| K900 | But as it stands, it is out of spec | 14:13:48 |
| toonn | I doubt I can convince anyone, mDNS seems to be rather universally despised. As long as I can get avahi to resolve subdomains for me, I'm happy. | 14:16:00 |
| toonn | magic_rb: Maybe I was naively hoping `pack . unpack` avoids reading the entire lazy bytestring into memory. I don't know if I elided the type declaration on purpose or just temporarily because I was changing things, probably the latter. | 14:24:21 |
| Jassuko | I would actually like mdns if it didn't have all those weird quirks on the implementations. -_- | 14:26:07 |
| toonn | It kinda feels like it was intentionally hamstrung to make it perpetually unpopular. | 14:27:46 |
| magic_rb | It does, going from a strict to lazy bytestring always incurs a copy unless the lazy bytestring is only one chunk i think. Then it may be able to do the conversion in place | 14:33:20 |
| toonn | magic_rb: I was thinking more that it might be read chunk by chunk, converted to the next however many Word8s and the same thing when packing, with fusion eliminating the intermediate list. | 14:34:55 |
| toonn | magic_rb: Also very possible I just didn't notice toStrict though. | 14:35:08 |
| magic_rb | https://hackage-content.haskell.org/package/bytestring-0.12.2.0/docs/src/Data.ByteString.Lazy.Internal.html#local-6989586621679118241 reading the impl here, it has a optimization for "just one chunk" | 14:36:13 |
| magic_rb | In that narrow case it should be O(1) and since you deal with up to 64 bytes, it should always be O(1) | 14:38:30 |
| magic_rb | Yeah the default chunk size is 32k - memory management overhead so your DNS packets fit as such lazy <-> strict is O(1) for you | 14:40:28 |
| toonn | Do you have a name on Codeberg? | 14:50:12 |
| toonn | For some Co-authored-by attribution ; ) | 14:50:51 |
| toonn | magic_rb: ^ | 16:12:49 |
| magic_rb | toonn: magic_rb magic_rb@redalder.org | 16:35:28 |
| magic_rb | * toonn: magic_rb <magic_rb@redalder.org> | 16:35:34 |
| toonn | Thanks : ) | 16:50:13 |
| 13 Sep 2025 |
|  Etherbloom [they/sie] {you&} 🐲🦊👿 changed their display name from Etherbloom [they/sie] {you&} 🐲🦊👿 to Etherbloom @MRMCD25 [they/sie] {you&} 🐲🦊👿. | 13:45:13 |
