libera/#commonlisp - IRC Chatlog
Search
14:15:15
Guest74
and I'm still not sure what this gets me over just defining a protocol in a protocol package which wouldn't have things like slots because it's a protocol and not an implementation.
14:15:17
beach
Guest74: Call it what you want. It is a separate package containing (so to speak) all code.
14:15:27
jackdaniel
and if you specify, that accessing slots is violating the protocol, then using slot-value disregarding the package should not happen
14:15:55
beach
Guest74: The concept is used informally in the literature. I formalized it a bit here: http://metamodular.com/protocol.pdf
14:19:30
mfiano
I just don't understand the reasoning for having only a package, without defgenerics for the implementation to conform to. It seems odd defining them in the implementation. I could see another parent node in between, but in the implementation? If this is only to avoid things like %-prefixing slots, I think it is quite an odd decision to conflate the API of the implementation with the
14:20:16
beach
Well, I need to get back to work. I am obviously not doing a good job today. Sorry about that.
14:22:56
beach
jackdaniel: There are no benefits to the client. I think the only benefit I mentioned today is that the -code (I wouldn't dare say -implementation again) package does not have to use any particular naming convention for slots.
14:23:09
Guest74
Well, I still don't see the benefit of separating the symbols of the protocol from the definition of the protocol, except it would satisfy some deep ocd corner of my brain.
14:23:26
jackdaniel
it doesn't have to do that either way (if we agree, that accessing slots directly is against the protocol)
14:26:38
jackdaniel
I'm not sure what I've already ageed upon, but since we all give up, then I'll make a coffee :)
14:26:51
Guest74
maybe it will make sense when I make a protocol for a purpose other than uniting objects from different libraries.
14:28:13
mfiano
Then why go to this much effort to try explaining an architecture that is useful to you?
14:29:38
mfiano
I wouldn't say that. I would just say that IRC is not the best medium where code examples and terms need to be cited as in a paper.
14:30:13
mfiano
I am still interested. I just don't think this forum is doing a good job of helping me understand.
14:45:04
jackdaniel
sometimes it is hard to set apart preferences from convictions; it is related to a set of cognitive biases, i.e anchoring and false consensus; I suppose these two are culpirts that often prevent seeing the other side argument
14:54:40
Guest74
now it's nice if you can just pass an object that contains your favourite dictionary, knows your preferred phonetic alphabet etc..., but what if you just want to change one setting for one call?
14:58:22
Guest74
or for vector graphics &key style (stroke (stroke style))(fill(fill style))(miter...
14:59:42
mfiano
If Uncle Bob is to be taken seriously, anything more than 2 or 3 arguments should be rolled into a "parameter object"
15:15:25
Guest74
I think there's a bot here that dislikes me asking what it thinks are stupid questions like 'why can't I name my function p(a|b)' and it kicks me.
16:07:45
mfiano
ACTION wonders how long it takes to use a real client with a real nick for someone very active on IRC
16:13:11
mfiano
Anyone with a Guest* nick, which are reserved for unregistered users on this network.
18:54:14
Bike
given that the web client you're using is presumably also not written in lisp, i don't see the relevance, though.
18:57:46
Guest74
bike: I'm not sure anybody uses those given that cl-irc is inflexible in the way it handles certain responses. Unless everybody just did what I did and change parts of cl-irc.
19:08:03
scymtym
my issue was that it does too much at a time (blocking) and too much on its own (via pre-registered handlers)
19:08:09
Guest74
but besides it not making space in some destructuring-binds for some extra stuff sent by libera I'm still using it. Just had to change stuff.
19:14:13
scymtym
i'll keep that in mind. maybe i should move the GUI code into a separate repository
19:16:30
Guest74
I don't currently use any of the other networks you support, but a more rounded out backend would be nice.
20:07:59
phoe
(also I see that this was already linked earlier, apologies for a very belated response)
20:11:56
pranavats
That Latex Parser is for LaTeX 2.0.9, has support for only a handful of LaTeX2e macros, and doesn't support the documentclass macro in particular. These can be fixed though.
20:15:06
scymtym
pranavats: https://github.com/scymtym/dpans-conversion/tree/online-lisp-meeting/code/parser (background: https://techfak.de/~jmoringe/presentation-dpans-conversion/slides.html ). but again: that is not a generic TeX parser