Have you tried this with Transit instead of EDN? From what I understand by all this Transit shouldn't have problems with spaces in keywords/strings as it doesn't print them in the same way, it's more of a marshaling format than a printer/reader, and you get the big upside of Transit being *way* faster than EDN.
Aside from that, I'd recommend just introducing a new type. It's easy to extend EDN, as you've mentioned, so why not just have a new type called IonSymbol that you use when talking to ION. Use defrecord, and you'll get equality semantics out of the box. On Thu, Aug 9, 2018 at 10:27 PM Didier <didi...@gmail.com> wrote: > Ion is a data serialization format from Amazon > http://amzn.github.io/ion-docs/ > > One lf its supported data type is symbol, defined as: Interned, Unicode > symbolic atoms (aka identifiers) > > I interact with systems that interchange their data using ION. My systems > use EDN though internally. Thus I need to convert from ION to EDN and back > in a lossless manner. > > If you give me ION with symbols in them, and I convert them to EDN, I need > be able to go back to the exact same ION symbol when converting my EDN back > to ION. > > If I make symbols string, I lose the type info, and would convert it back > to an ION string. This wouod break my clients. > > I thought I could get away not having to build a custom representation for > them in Clojure and EDN, and piggyback on keywords. > > It now seems to me like I can still piggyback on keywords when in Clojure, > since `keyword` and clojure.lang.Keyword appear to support full unicode. > > But it seems EDN does not support full unicode serialization for them. So > I've extended the edn serialization and deserialization so that it does. > > I did so by overriding print-method for keywords so that when it prints in > an unreadable way, it instead outputs a custom literal #myproject/keyword > "namespace/name". And I've added a reader for this which uses keyword to > get a Clojure keyword back. > > Does any of this seem unsound to anyone? > > -- > 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. > -- “One of the main causes of the fall of the Roman Empire was that–lacking zero–they had no way to indicate successful termination of their C programs.” (Robert Firth) -- 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.
