Answers below are my perspective. I have no doubt that Rich would give you better ones. :)
On Tuesday, November 29, 2016 at 1:55:59 PM UTC-6, Leon Grapenthin wrote: > > While I can follow "we have namespaced names" I find it difficult to > follow "too much stuff onto vars and var meta". I feel this is what OP is > asking for. > > - What is meant by "too much"? > More than we have now. > When is var metadata "too much"? > Var meta gets compiled into the bytecode. This affects compiled code size and more importantly loading time. Also related to load time are vars which are loaded and initialized when a namespace is loaded. These are real costs that affect startup and load times. > - What problem can/did result from "too much"? > Load time. Compiled code size. Abusing vars as "the one place to hang everything". > - Why is this assumed limitation always stated as primary motivation? Are > there not other reasons like enabling non var-tied specs like for keywords? > I don't understand the question. When you have namespaced names, you can create many registries for different purposes and stop using vars as The Registry. There may be other things that could be "registries" as well, rather than being tied only to vars. > - Have s/fdefs been considered for vars? What is the reason/intended use > for s/fdefs in the global registry? > Same reasons, same answers. > > Kind regards, > Leon > > On Tuesday, November 29, 2016 at 5:16:36 PM UTC+1, Alex Miller wrote: >> >> At the risk of repeating what you already said, we are pushing too much >> stuff onto vars and var meta. We have namespaced names so there's no reason >> to push more stuff into vars and we can use an independent registry. >> >> >> On Tuesday, November 29, 2016 at 8:57:47 AM UTC-6, kovasb wrote: >>> >>> Spec is surprisingly easy to grok given how much it does. >>> >>> s/def jumped out at me as an out-of-the-box choice, that I could not >>> immediately rationalize. >>> >>> So I'm wondering: why not just use standard def? What does one gain/lose? >>> >>> This is not just an academic question. Lots of people like me look at >>> Clojure's APIs as an example to follow, so now I'm wondering when I should >>> make the same choice. >>> >>> I asked Rich about this at the Lisp NYC meetup, the response was to the >>> effect of "vars are already overloaded, lets use a separate database", >>> which makes sense but the implications of that are not clicking. >>> >>> >>> >>> >>> -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en --- You received this message because you are subscribed to the Google Groups "Clojure" group. To unsubscribe from this group and stop receiving emails from it, send an email to clojure+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
