rearrange the test code and add some notes on postgres

TODO

@@ -1,85 +1,100 @@
 
 next release:
-quoted identifiers: implement as a dot operator
-review tests to copy from hssqlppp
+lateral since it is easy. I think it is effectively just a prefix
+   operator on tablerefs (can appear like this from lateral a, from
+   lateral a,b, from a, lateral b, from a natural join lateral b
+   (although all of these are sematically garbage, this is the
+   valid syntax)
+more dots: implement as dot operator
+
+row ctor: row(a,b) is fine, but also when there is 2 or more elements,
+   the word row can be omitted: (a,b)
+more symbolic operators, array access a[5]? don't think this is
+   standard sql, if not, leave for now. There is something about
+   arrays in sql:2008
+
 window frames and named windows
-dialect framework
-try to implement fixity without the hse hack
-position annotation?
-row ctor
-more symbolic operators, array access a[5]?
-review abstract syntax (e.g. combine App with SpecialOp?)
-more dots
+
+
+review tests to copy from hssqlppp
+
 order by nulls first/last
 extend case
-group by extensions
+group by extensions. Question: some of the syntax can be represented
+   by app and row ctor, should this be reused or new syntax created
+   (the standard has special syntax for cube and rollup).
 table, values
-collate? -> postfix operator which binds very  tightly:
+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
+select a from t order by a collate 'C': add to order by syntax, one
+   collation per column
 function table reference
-much more table reference tests
+much more table reference tests, for joins and aliases etc.
 ansi standard versions of limit and offset
-sql server top
+
+OFFSET start { ROW | ROWS }
+FETCH { FIRST | NEXT } [ count ] { ROW | ROWS } ONLY
+-> + fix the abstract syntax to match this instead of postgres
+(keep the postgres syntax version parser)
+in the postgresql docs, the start and count must be in parens unless
+   they are a single integer
+
++ sql server top syntax
+
+quoted identifiers and proper character sets for identifiers
+
+run through postgres docs and add example sql as tests
+review internal sql collection for more syntax/tests
+all ansi sql operators
+
+escapes in string literals
+
+
+----
+
 add to website: pretty printed tpch, maybe other queries as
    demonstration
-demo: convert tpch to sql server syntax
-review internal sql collection for more syntax/tests
-run through postgres docs and add example sql as tests
+demo: convert tpch to sql server syntax exe processor
+
+----
+
+dialect framework
+try to implement fixity without the hse hack
+position annotation?
+review abstract syntax (e.g. combine App with SpecialOp?)
+
 
 ----
 
 Later general tasks:
 
+run through other manuals for example queries and features: sql in a
+   nutshell, sql guide, sql reference guide, sql standard, sql server
+   manual, oracle manual, teradata manual + re-through postgresql
+   manual and make notes in each case of all syntax and which isn't
+   currently supported also.
+
 check the order of exports, imports and functions/cases in the files
 fix up the import namespaces/explicit names nicely
 
 do some tests for parse errors?
 
-dialect switching
-
 left factor parsing code in remaining places
 
-reimplement the fixity thing natively
-
-position annotation?
-
 quasi quotes?
 
 = sql support
 
-scalar function syntax:
+placeholder/positional arg
 
-  review all ansi sql operators
-  placeholder/positional arg
-
-other missing operators
-  row constructors -> needed for stuff like
-  'where (a,b) = any (select a,b from t)'
-
-
-review allowed identifier syntax
-  add quoted identifiers
-more dots in identifiers
-order by nulls first/last
-extend case
-escapes in string literals
 full number literals -> other bases?
-group by (), grouping sets(), cube, rollup
-lateral
-named windows
-table, values
 apply, pivot
-collate
 within group aggregate syntax
 
-support the ansi version for limit and offset, plus review other
-   dialects
-
 try to do full review of sql2003 query syntax
 
 make ansi dialect which only supports ansi sql. Maybe there is a use
@@ -94,3 +109,4 @@ mysql?
 db2?
 
 maybe later: other dml
+  insert, update, delete, truncate, merge + set, show?