The reader is mostly complete. It follows the basic outline of LispReader.java, except in a non-object-oriented style and with some simplifications.

The code is all in the reader.cljs file. The primary function isread, which takes an instance of thePushbackReader protocol. An implementation of this protocol is provided,StringPushbackReader, and may be constructed easily using thepush-back-reader function, which takes a string and converts it to aStringPushBackReader.

There are a number of known issues/limitations.

• There are some cases where a map may be incorrectly read as a set. An example that does this consistently is the form(1 2 3 {:a 1 :b 2})
• Does not support hexadecimal or custom-radix numeric literals (e.g, 0xA3 or 8r3). These cases will throw an "Invalid Number Format" error. Octal numbers starting with a 0 are parsed as base-10 numbers. This is due to a difference in the Javascript and Java regex engines: the regex used in both cases is identical. The javascript one needs to be tested and modified before hex, octal and custom-radix numbers will work.
• The performance of the StringPushbackReader is currently poor, since it actually modifying a string on each read/unread. It would be much faster if it were rewritten to work by indexing into a constant string, with a separate small buffer to hold pushbacks.