[00:06] <Voldenet> .raku isn't supposed to be serialization at all

[00:07] <Voldenet> it can be used like that, but it'd require EVAL, which pretty much doesn't work

[00:08] <Voldenet> in the context of serialization

[00:08] <Voldenet> I'd rather not deserialize things that can execute random code :>

[00:14] *** rmv left
[00:31] <wayland76> ...and that's what I meant about security :) 

[01:31] *** topnep left
[01:32] *** topnep joined
[01:40] *** kylese left
[01:40] *** hulk joined
[01:48] *** leppard left
[02:15] *** hulk left
[02:15] *** kylese joined
[02:25] *** tejr left
[02:30] *** tejr joined
[02:35] *** ShimmerFairy left
[02:43] *** ShimmerFairy joined
[03:11] *** guifa_ left
[03:35] *** banna joined
[03:36] *** topnep left
[03:36] *** topnep joined
[04:15] <japhb> If you want to do round-trip serialization, CBOR::Simple is your friend.

[05:16] *** guifa left
[05:19] *** guifa joined
[06:15] *** Sgeo left
[06:49] *** abraxxa joined
[07:45] *** rmv joined
[07:45] *** rmv left
[07:45] *** rmv joined
[07:50] *** rmv left
[08:17] *** rmv joined
[08:21] *** rmv left
[09:07] *** wayland76 left
[09:07] *** wayland joined
[10:16] *** rmv joined
[10:16] *** rmv left
[10:16] *** rmv joined
[10:33] *** apogee_ntv joined
[10:46] *** abraxxa left
[10:52] *** guifa_ joined
[11:12] *** sergot joined
[11:23] *** guifa_ left
[12:32] *** topnep left
[12:58] <tbrowder> .seen El_Che

[12:58] <tellable6> tbrowder, I saw El_Che 2026-07-07T10:15:07Z in #raku: <El_Che> normally it's done fast

[12:59] <tbrowder> El_Che: is the rakudo-pkg working for debian trixie?

[13:28] *** acidsys left
[13:46] *** acidsys joined
[14:00] *** Sgeo joined
[16:55] <ab5tract> Can we create another grammar braid (a "data" braid) that implements a very slim subset of grammar rules from the regular parser?

[16:59] <Voldenet> well – the biggest problem with serialization is if I want to serialize things like Proc::Async inside some bigger object

[17:00] <Voldenet> obviously, serialization can get cmdline and deserialization should start it back, but you need to be able to specify serializers

[17:01] <Voldenet> and sometimes it's not known whether what is secure in the specific context of serialization (one application can have context that stores the same thing to persist in db, other instances or clients)

[17:02] <Voldenet> s/whether//

[17:02] <ab5tract> I think we are talking about different things. I am talking about a safe-to-EVAL DDL (like a JSON but with Raku syntax)

[17:02] <Voldenet> aye, like you configure running app and dump config.raku file

[17:02] <ab5tract> What you are talking about reminds me of Storable, and so I would expect similar limitations and security concerns

[17:03] <ab5tract> which is to say: there be yikes yonder, matey :)

[17:05] <ab5tract> I would say the issue is mostly resolved by leaving it to the applications to create the serialization/deserialization logic. This could be via a library that is shared amongst applications.

[17:05] <ab5tract> So maybe we are saying the same thing after all? ;)

[17:06] <Voldenet> if you can share the library

[17:06] *** guifa left
[17:06] *** guifa_ joined
[17:07] <Voldenet> in sql there's built-in json/xml, same in browsers

[17:08] <Voldenet> so in the end, JSON::Fast paired with `method to-db-dto() returns Hash { … }` is the way

[17:08] <Voldenet> at the moment

[17:09] <Voldenet> I'm tempted to try adjusting JSON::Fast to more oop approach

[17:10] <Voldenet> it sounds like a bit of work, but since tokens are parsed already it could be not that horrible

[17:11] <ab5tract> Voldenet: The thing is that "dump a config to disk" to me implies that the application knows what information to save (some of which might be in private attributes) as config and then how to read that config back and re-construct the various objects that are being configured

[17:14] <ab5tract> that said, a simple role with some sane defaults and an override method could probably handle a large majority of cases

[17:23] <Voldenet> yeah, "serialize/deserialize" operation (saying what is temporary and what is persistent) has to be somewhere in most cases

[17:31] *** human-blip left
[17:33] *** human-blip joined
[17:38] *** abraxxa joined
[18:10] *** abraxxa left
[18:22] *** leppard joined
[18:36] *** lichtkind joined
[18:41] *** rmv left
[18:51] *** rmv joined
[19:04] *** rmv left
[20:04] *** jdv left
[20:06] *** rmv joined
[20:06] *** rmv left
[20:06] *** rmv joined
[21:02] *** Pixi` joined
[21:06] *** Pixi left
[21:18] *** guifa joined
[21:19] *** rmv left
[21:29] *** shareable6 left
[21:32] *** shareable6 joined
[21:41] *** guifa left
[23:18] *** rmv joined
[23:18] *** rmv left
[23:18] *** rmv joined
[23:23] *** rmv left
