This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2016-11-08
Channels
- # aws (3)
- # bangalore-clj (1)
- # beginners (47)
- # boot (137)
- # cider (1)
- # cljs-dev (67)
- # cljsrn (7)
- # clojure (122)
- # clojure-argentina (5)
- # clojure-berlin (4)
- # clojure-czech (12)
- # clojure-france (27)
- # clojure-italy (9)
- # clojure-russia (88)
- # clojure-spec (44)
- # clojure-uk (157)
- # clojurebridge (2)
- # clojurescript (236)
- # datomic (5)
- # devcards (3)
- # dirac (23)
- # emacs (13)
- # hoplon (29)
- # incanter (1)
- # leiningen (41)
- # microservices (1)
- # off-topic (78)
- # om (145)
- # onyx (13)
- # parinfer (8)
- # pedestal (4)
- # planck (15)
- # protorepl (1)
- # re-frame (72)
- # reagent (25)
- # ring (2)
- # specter (23)
- # test-check (9)
- # untangled (106)
- # vim (8)
- # yada (1)
@thheller: I don't see what the problem is with respect to the current behavior
It doesn't change anything, just makes it customizable...
@anmonteiro yeah but why make it customizable
writing faulty analysis data that cannot be re-created to the same state helps no one
@thheller well people may want the analysis cache to be written out in EDN even if Transit is in the classpath
but the underlying problem is still the same ... there is un-readable data in the cache
right but that's solved by toggling the switch to :edn
, isn't it?
then the analysis is written and read as EDN 🙂
right now it doesn't really matter since nobody is using this data, but it is still an error
how come nobody is using this data?
Transit support has been there for a while
cache wants to write arglists .. say (defn my-fn [{:keys [thing] :or {thing #js {}}])
the problem is not use edn or transit, the problem is write in analysis cache data that is useless. When transit is used, an exception is raised, when edn is used, it is silently written to cache (but it useless, becasue when that cache is readed again, previously written bad data it is useless).
and @thheller proposal is just does not write analysis cache when it is not posible to encode valid data... (I'm right?)
oh OK I get it
@thheller so IMHO you're conflating things...
because the problem you're referring to exists today, and has existed for a while
so why mention it as part of this ticket at all?!
I think they are two different things
people should be able to choose to write the cache as edn vs transit even if there's the EDN problem
@anmonteiro because the ticket introduces an additional compiler option that is not useful
hmm @thheller I agree here with @anmonteiro , the option is useful
solving the second issue will allow use transit in all cases, because cache that can't be encoded will just gracefully not written to cache instead crash...
@anmonteiro @niwinz opened http://dev.clojure.org/jira/browse/CLJS-1843 so the option can be discussion separately
@dnolen before I attach it to JIRA, how's this one based on your feedback yesterday? https://github.com/anmonteiro/clojurescript/commit/2b7e40705bcbd92ed3ee1e607ac10c4dad4fdc2d
- added validation for the :cache-analysis-format
option
- explicit inference of the cache file format via the new function cache-analysis-ext
@anmonteiro looks good enough to me
and we didn't need to change the API
@dnolen attaching it to JIRA then, thanks