3 ms·
It's basically a kind of step back to the Lisp of the 60s and seventies. Unfortunately I don't think this approach scales well. For example I prefer to look in
by lispm 13y ago
It's basically a kind of step back to the Lisp of the 60s and seventies. Unfortunately I don't think this approach scales well. For example I prefer to look in a debugger or inspector at an object and not at some other map.
- adrianm 13y agoI'm going to be honest - you are coming off as very contrarian for the sake of being contrarian. You seem to be consistently responding to my comments in this thread with the equivalent of "You're wrong, because I like to do things this way." That isn't an argument against any point I've made. Please show me some code and give me some context that will make me rethink my assumptions and arguments if you feel the need to point out my flawed thinking.
- lispm 13y agoNo, plain maps are nothing more like plain assoc lists (and similar stuff) in plain old Lisp. We were modelling key/value data structures like that from the earliest Lisp. Sure its easy to use. Up to a point. Lisp had an evolution. Things changed. I've seen quite a bit larger Lisp software over the years and the ones not using explicit classes turned out to be harder to maintain.
- weavejester 13y agoI'd contend that it's the one approach we know does scale well. The largest systems we've built, such as the web, are designed around bare data structures, rather than objects.
- dragonwriter 13y agoNo, large heterogenous systems like the web are built on serialization formats that different components within that same heterogenous system can (and do) interpret as either bare data structures, objects, or just unstructured streams of text (or binary data.)
- weavejester 13y agoSerialised data is still data - a sequence of bytes representing a JSON object is still semantically the same when it's decoded into a hash map. The information content remains the same.