update TODO

TODO

@@ -1,7 +1,8 @@
 = next release
 
-Most important goal is to replace the fixity code and fix all the
+Most important goal is to replace the fixity code and fix all the bugs
-   bugs here.
+   here. Could also review parens and fixity at query expr level, and
+   in tablerefs
 
 == docs
 
@@ -23,27 +24,13 @@ review internal sql collection for more syntax/tests
 
 == features
 
-collate? -> postfix operator which binds very tightly:
+parse '?'
-a < 'foo' collate 'C'
-->
-Op "<" [Iden "a", SpecialOp "collate" [StringLit 'foo', StringLit
-   'C']]
-  also postfix in order by:
-select a from t order by a collate 'C': add to order by syntax, one
-   collation per column
-have to do fixity for this to work
-
-all ansi string literal syntax?
-
-check ansi standard for operators (keywords and symbols)
 
 == other
 
-change any/some/all to be proper infix operators like in
+change any/some/all to be proper infix operators like in ??
 
 review syntax to replace maybe and bool with better ctors
-maybe review some of the dodgy ast names like orderfield and inthing
-
 
 
 = Later general tasks:
@@ -86,6 +73,23 @@ ast checker: checks the ast represents valid syntax, the parser
 
 = sql support
 
+collate? -> postfix operator which binds very tightly:
+a < 'foo' collate 'C'
+->
+Op "<" [Iden "a", SpecialOp "collate" [StringLit 'foo', StringLit
+   'C']]
+  also postfix in order by:
+select a from t order by a collate 'C': add to order by syntax, one
+   collation per column
+and on group by
+a collation in ansi sql is a optional qualifier plus identifier, not a
+   string
+have to do fixity for this to work
+
+all ansi string literal syntax?
+
+check ansi standard for operators (keywords and symbols)
+
 proper character sets for identifiers, escapes, etc.
 
 placeholder/positional arg
@@ -99,6 +103,50 @@ try to do full review of sql2003 query syntax
 make ansi dialect which only supports ansi sql. Maybe there is a use
    for a separate SQL 92, 99 and 2003 dialects?
 
+string literals:
+N'
+B'
+X'
+U&' ?
+looks like two string literals separated by any whitespace are
+   merged, which isn't what the postgres manual claims is the standard
+plus something introducer, character set thing
+union join?
+corresponding by
+from in substring is not optional in ansi
+the keyword from is optional in trim iff there is neither
+   leading,etc. and explicit char
+extract can accept 'timezone hour' and 'timezone minute' for the
+   extraction field. Could hardcode the complete allowed list in the
+   parser?
+bunch of timezone keyword stuff
+interval a to b
+array specification int array or int array[5] - type name
+typenames: scope, multiset,row type, ref type
+unique subquery -> returns true if all the rows in the subquery are
+   distinct
+match predicate <row value constructor> MATCH [ UNIQUE ] [ PARTIAL |
+   FULL ] <table subquery>
+clob type: CLOB(4M) - M suffix on the precision
+array element reference a[4]
+regular expression substring function
+between symmetric
+match predicate
+unicode delimited identifier
+character set postfix type op for string types + collate on string types
+next value for
+treat as
+generalized invocation?
+position using
+multiset stuff
+table sample
+unnest
+only
+window frame exclusion: complete window support?
+select t.* as (a,b,c) from t ??
+search or cycle clause for ctes
+normalized
+
 other dialect targets:
 postgres
 oracle