Обсуждение: patch: Add JSON datatype to PostgreSQL (GSoC, WIP)

This is a work-in-progress patch of my GSoC project: Add JSON datatype
to PostgreSQL.  It provides the following:

 * JSON datatype: A TEXT-like datatype for holding JSON-formatted
text.  Although the JSON RFC decrees that a JSON text be an "object or
array" (meaning '"hello"' is considered invalid JSON text), this
datatype lets you store any JSON "value" (meaning '"hello"'::JSON is
allowed).
 * Validation: Content is validated when a JSON datum is constructed,
but JSON validation can also be done programmatically with the
json_validate() function.
 * Conversion to/from JSON for basic types.  Conversion functions are
needed because casting will not unwrap JSON-encoded values.  For
instance, json('"string"')::text is '"string"', while
from_json('"string"') is 'string'.  Also, to_json can convert
PostgreSQL arrays to JSON arrays, providing a nice option for dealing
with arrays client-side.  from_json currently can't handle JSON
arrays/objects yet (how they should act is rather unclear to me,
except when array dimensions and element type are consistent).
 * Retrieving/setting values in a JSON node (via selectors very
similar to, but not 100% like, JSONPath as described at
http://goessner.net/articles/JsonPath/ ).
 * Miscellaneous functions json_condense and json_type.

This is a patch against CVS HEAD.  This module compiles, installs, and
passes all 8 tests successfully on my Ubuntu 9.10 system.  It is
covered pretty decently with regression tests.  It also has SGML
documentation (the generated HTML is attached for convenience).

Although I am aware of many problems in this patch, I'd like to put it
out sooner rather than later so it can get plenty of peer review.
Problems I'm aware of include:
 * Probably won't work properly when the encoding (client or server?)
is not UTF-8.  When encoding (e.g. with json_condense), it should (but
doesn't) use \uXXXX escapes for characters the target encoding doesn't
support.
 * json.c is rather autarkic.  It has its own string buffer system
(rather than using StringInfo) and UTF-8 validator (rather than using
pg_verify_mbstr_len(?) ).
 * Some functions/structures are named suggestively, as if they belong
to (and would be nice to have in) PostgreSQL's utility libraries.
They are:
   - TypeInfo, initTypeInfo, and getTypeInfo: A less cumbersome
wrapper around get_type_io_data.
   - FN_EXTRA and FN_EXTRA_SZ: Macros to make working with
fcinfo->flinfo->fn_extra easier.
   - enumLabelToOid: Look up the Oid of an enum label; needed to
return an enum that isn't built-in.
   - utf8_substring: Extract a range of UTF-8 characters out of a UTF-8 string.
 * Capitalization and function arrangement are rather inconsistent.
Braces are K&R-style.
 * json_cleanup and company aren't even used.
 * The sql/json.sql test case should be broken into more files.

P.S. The patch is gzipped because it expands to 2.6 megabytes.


Joey Adams

On Fri, Jul 23, 2010 at 2:18 AM, Joseph Adams
<joeyadams3.14159@gmail.com> wrote:
> This is a work-in-progress patch of my GSoC project: Add JSON datatype
> to PostgreSQL.  It provides the following:
>
>  * JSON datatype: A TEXT-like datatype for holding JSON-formatted
> text.  Although the JSON RFC decrees that a JSON text be an "object or
> array" (meaning '"hello"' is considered invalid JSON text), this
> datatype lets you store any JSON "value" (meaning '"hello"'::JSON is
> allowed).
>  * Validation: Content is validated when a JSON datum is constructed,
> but JSON validation can also be done programmatically with the
> json_validate() function.
>  * Conversion to/from JSON for basic types.  Conversion functions are
> needed because casting will not unwrap JSON-encoded values.  For
> instance, json('"string"')::text is '"string"', while
> from_json('"string"') is 'string'.  Also, to_json can convert
> PostgreSQL arrays to JSON arrays, providing a nice option for dealing
> with arrays client-side.  from_json currently can't handle JSON
> arrays/objects yet (how they should act is rather unclear to me,
> except when array dimensions and element type are consistent).
>  * Retrieving/setting values in a JSON node (via selectors very
> similar to, but not 100% like, JSONPath as described at
> http://goessner.net/articles/JsonPath/ ).
>  * Miscellaneous functions json_condense and json_type.
>
> This is a patch against CVS HEAD.  This module compiles, installs, and
> passes all 8 tests successfully on my Ubuntu 9.10 system.  It is
> covered pretty decently with regression tests.  It also has SGML
> documentation (the generated HTML is attached for convenience).
>
> Although I am aware of many problems in this patch, I'd like to put it
> out sooner rather than later so it can get plenty of peer review.
> Problems I'm aware of include:
>  * Probably won't work properly when the encoding (client or server?)
> is not UTF-8.  When encoding (e.g. with json_condense), it should (but
> doesn't) use \uXXXX escapes for characters the target encoding doesn't
> support.
>  * json.c is rather autarkic.  It has its own string buffer system
> (rather than using StringInfo) and UTF-8 validator (rather than using
> pg_verify_mbstr_len(?) ).
>  * Some functions/structures are named suggestively, as if they belong
> to (and would be nice to have in) PostgreSQL's utility libraries.
> They are:
>   - TypeInfo, initTypeInfo, and getTypeInfo: A less cumbersome
> wrapper around get_type_io_data.
>   - FN_EXTRA and FN_EXTRA_SZ: Macros to make working with
> fcinfo->flinfo->fn_extra easier.
>   - enumLabelToOid: Look up the Oid of an enum label; needed to
> return an enum that isn't built-in.
>   - utf8_substring: Extract a range of UTF-8 characters out of a UTF-8 string.
>  * Capitalization and function arrangement are rather inconsistent.
> Braces are K&R-style.
>  * json_cleanup and company aren't even used.
>  * The sql/json.sql test case should be broken into more files.
>
> P.S. The patch is gzipped because it expands to 2.6 megabytes.

Some technical points about the submission:

- If you want your coded to be included in PostgreSQL, you need to put
the same license and attribution on it that we use elsewhere.
Generally that looks sorta like this:

/*-------------------------------------------------------------------------** tablecmds.c*        Commands for creating
andaltering table structures and settings** Portions Copyright (c) 1996-2010, PostgreSQL Global Development Group*
PortionsCopyright (c) 1994, Regents of the University of California*** IDENTIFICATION*
$PostgreSQL$**-------------------------------------------------------------------------*/

You should have this header on each file, both .c and .h.

- The reason why this patch is 2.6MB is because it has 2.4MB of tests.I think you should probably pick out the most
useful10% or so, and 
drop the rest.

- I was under the impression that we wanted EXPLAIN (FORMAT JSON) to
return type json, but that's obviously not going to be possible if all
of this is contrib.  We could (a) move it all into core, (b) move the
type itself and its input and output functions into core and leave the
rest in contrib [or something along those lines], or (c) give up using
it as the return type for EXPLAIN (FORMAT JSON).

- You need to comply with the project coding standards.  Thus:

static void
foo()
{   exit(1);
}

Not:

static void foo()
{   exit(1);
}

You should have at least one blank line between every pair of
functions.  You should use uncuddled curly braces.  Your commenting
style doesn't match the project standard.  We prefer explicit NULL
tests over if (!foo).  Basically, you should run pgindent on your
code, and also read this:

http://www.postgresql.org/docs/8.4/static/source.html

I don't think this is going to fly either:

/* Declare and initialize a String with the given name. */
#define String(name) String name = NewString()
#define NewString() {{NULL, 0, 0}}

That's just too much magic.  We want people to be able to read this
code and easily tell what's going on...  spelling a few things out
long hand is OK, good, even.

- You have boatloads of functions in here with no comments whose
functions is entirely non-self-evident.  You may or may not need to
rename some of them, but you definitely need to write some comments.

- elog() must be used except for "can't happen" situations.  Compare
the way numeric_in() throws an error message versus json_in().

- You have a number of very short convenience functions which don't
seem warranted.  For example, build_array_string() is a 2-line
function that is called once.  And all the string_append,
string_append_range, string_append_char stuff is duplicating
functionality we already have in appendStringInfoStr,
appendBinaryStringInfo, appendStringInfoChar.

In short, my first impression of this patch is that there's a lot of
good stuff in here, but you need to stop adding features NOW and put A
LOT of work into getting this into a form that the community can
accept.  Otherwise, this is likely going to die on the vine, which
would be a shame because a lot of it seems pretty cool.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

On Fri, Jul 23, 2010 at 2:34 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> - elog() must be used except for "can't happen" situations.  Compare
> the way numeric_in() throws an error message versus json_in().

Er... that should have said "elog() must NOT be used except for can't
happen situations".

Also, I was just looking at json_delete().  While the existing coding
there is kind of fun, I think this can be written more
straightforwardly by doing something like this (not tested):
   while (1)   {       while (is_internal(node) && node->v.children.head)           node = node->v.children.head;
if(node->next)           next = node->next;       else if (node->parent)           next = node->parent;       else
    break;       free_node(node);       node = next;   } 

That gets rid of all of the gotos and one of the local variables and,
at least IMO, is easier to understand...  though it would be even
better still if you also added a comment saying something like "We do
a depth-first, left-to-right traversal of the tree, freeing nodes as
we go.  We need not bother clearing any of the pointers, because the
traversal order is such that we're never in danger of revisiting a
node we've already freed."

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

Update: I'm in the middle of cleaning up the JSON code (
http://git.postgresql.org/gitweb?p=json-datatype.git;a=summary if you
want to see the very latest ), so I haven't addressed all of the major
problems with it yet.

On Fri, Jul 23, 2010 at 2:34 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> - I was under the impression that we wanted EXPLAIN (FORMAT JSON) to
> return type json, but that's obviously not going to be possible if all
> of this is contrib.  We could (a) move it all into core, (b) move the
> type itself and its input and output functions into core and leave the
> rest in contrib [or something along those lines], or (c) give up using
> it as the return type for EXPLAIN (FORMAT JSON).

I've been developing it as a contrib module because:
* I'd imagine it's easier than developing it as a built-in datatype
right away (e.g. editing a .sql.in file versus editing pg_type.h ).
* As a module, it has PGXS support, so people can try it out right
away rather than having to recompile PostgreSQL.

I, for one, think it would be great if the JSON datatype were all in
core :-)  However, if and how much JSON code should go into core is up
for discussion.  Thoughts, anyone?

A particularly useful aspect of the JSON support is the ability to
convert PostgreSQL arrays to JSON arrays (using to_json ), as there
currently isn't any streamlined way to parse arrays in the PostgreSQL
format client-side (that I know of).


Joey Adams

On Sat, Jul 24, 2010 at 06:57:18PM -0400, Joseph Adams wrote:
> A particularly useful aspect of the JSON support is the ability to
> convert PostgreSQL arrays to JSON arrays (using to_json ), as there
> currently isn't any streamlined way to parse arrays in the PostgreSQL
> format client-side (that I know of).
I really would like to address the latter issue some day. I don't know
how many broken implementations I have seen in my, not that long, time
using pg, but it sure is 10+. I also knowingly have written dumbed
down versions.

Andres

> I, for one, think it would be great if the JSON datatype were all in
> core :-)  However, if and how much JSON code should go into core >is up for discussion.  Thoughts, anyone?
>
in my opinion: As soon as possible. Spinning PostgreSQL as the
Ajax-enabled-database has many great uses.

Harald

--
GHUM Harald Massa
persuadere et programmare
Harald Armin Massa
Spielberger Straße 49
70435 Stuttgart
0173/9409607
no fx, no carrier pigeon
-
Using PostgreSQL is mostly about sleeping well at night.

Updated JSON datatype patch.  It cleans up the major problems that
have been discussed, and it's very close to being commit-worthy (I
think).  The major issues as I see them are:

 * Contains several utility functions that may be useful in general.
They are all in util.c / util.h
 * It's still a contrib module
 * No json_agg or json_object functions for constructing arrays / objects

The utility functions and their potential to collide with themselves
in the future is the main problem with this patch.  Of course, this
problem could be sidestepped simply by namespacifying them (prepending
json_ to all the function names).  I would like some thoughts and
opinions about the design and usefulness of the utility code.

An overview, along with my thoughts, of the utility functions:

FN_EXTRA, FN_EXTRA_ALLOC, FN_MCXT macros
 * Useful-ometer: ()--------------------o
 * Rationale: Using fcinfo->flinfo->fn_extra takes a lot of
boilerplate.  These macros help cut down the boilerplate, and the
comment explains what fn_extra is all about.

TypeInfo structure and getTypeInfo function
 * Useful-ometer: ()---------------------------o
 * Rationale: The get_type_io_data "six-fer" function is very
cumbersome to use, since one has to declare all the output variables.
The getTypeInfo puts the results in a structure.  It also performs the
fmgr_info_cxt step, which is a step done after every usage of
get_type_io_data in the PostgreSQL code.

getEnumLabelOids
 * Useful-ometer: ()-----------------------------------o
 * Rationale: There is currently no streamlined way to return a custom
enum value from a PostgreSQL function written in C.  This function
performs a batch lookup of enum OIDs, which can then be cached with
fn_extra.  This should be reasonably efficient, and it's quite elegant
to use (see json_op.c for an example).

UTF-8 functions:
    utf8_substring
    utf8_decode_char
        (there's a patch in the works for a utf8_to_unicode function
         which does the same thing as this function)
    utf8_validate (variant of pg_verify_mbstr(PG_UTF8, str, length, true)
that allows '\0' characters)
    server_to_utf8
    utf8_to_server
    text_to_utf8_cstring
    utf8_cstring_to_text
    utf8_cstring_to_text_with_len
 * Useful-ometer: ()-------o
 * Rationale: The JSON code primarily operates in UTF-8 rather than
the server encoding because it needs to deal with Unicode escapes, and
there isn't an efficient way to encode/decode Unicode codepoints
to/from the server encoding.  These functions make it easy to perform
encoding conversions needed for the JSON datatype.  However, they're
not very useful when operating solely in the server encoding, hence
the low usefulometric reading.

As for the JSON datatype support itself, nobody has come out against
making JSON a core datatype rather than a contrib module, so I will
proceed with making it one.  I guess this would involve adding entries
to pg_type.h and pg_proc.h .  Where would I put the rest of the code?
I guess json_io.c and json_op.c (the PG_FUNCTION_ARGS functions) would
become json.c in src/backend/utils/adt .  Where would json.c and
jsonpath.c (JSON encoding/decoding functions and JSONPath
implementation) go?

Are there any other issues with the JSON code I didn't spot?

Thanks,


Joey Adams

Excerpts from Joseph Adams's message of mar ago 10 04:03:43 -0400 2010:

> An overview, along with my thoughts, of the utility functions:
> 
> FN_EXTRA, FN_EXTRA_ALLOC, FN_MCXT macros
>  * Useful-ometer: ()--------------------o
> 
> TypeInfo structure and getTypeInfo function
>  * Useful-ometer: ()---------------------------o
>
> getEnumLabelOids
>  * Useful-ometer: ()-----------------------------------o

I think this kind of thing could be stripped from the patch and
submitted separately; they would presumably see a quick review and
commit if they are small and useful (particularly if you can show a
decrease of code verbosity by switching other uses in the existing
code).

The advantage is you don't have to keep arguing for their usefulness in
the JSON patch; and if they turn out to be rejected, they won't cause
the JSON patch to be rejected as a whole.

-- 
Álvaro Herrera <alvherre@commandprompt.com>
The PostgreSQL Company - Command Prompt, Inc.
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

Alvaro Herrera <alvherre@commandprompt.com> writes:
> Excerpts from Joseph Adams's message of mar ago 10 04:03:43 -0400 2010:
>> An overview, along with my thoughts, of the utility functions:
>> 
>> FN_EXTRA, FN_EXTRA_ALLOC, FN_MCXT macros
>> * Useful-ometer: ()--------------------o
>> 
>> TypeInfo structure and getTypeInfo function
>> * Useful-ometer: ()---------------------------o
>> 
>> getEnumLabelOids
>> * Useful-ometer: ()-----------------------------------o

> I think this kind of thing could be stripped from the patch and
> submitted separately;

+1.  It's easier all around if a patch does just one thing.  Code
refactoring and feature addition, in particular, are easier to review
separately.
        regards, tom lane

On Sat, 2010-07-24 at 18:57 -0400, Joseph Adams wrote:

> I've been developing it as a contrib module because:
> * I'd imagine it's easier than developing it as a built-in datatype
> right away (e.g. editing a .sql.in file versus editing pg_type.h ).
> * As a module, it has PGXS support, so people can try it out right
> away rather than having to recompile PostgreSQL.
> 
> I, for one, think it would be great if the JSON datatype were all in
> core :-)  However, if and how much JSON code should go into core is up
> for discussion.  Thoughts, anyone?

As a GSoC piece of work, doing it as a contrib module gives an
immediately useful deliverable. Good plan.

Once that is available, we can then get some feedback on it and include
it as an in-core datatype later in the 9.1 cycle.

So lets do both: contrib and in-core.

-- Simon Riggs           www.2ndQuadrant.comPostgreSQL Development, 24x7 Support, Training and Services

On Sat, Jul 24, 2010 at 06:57:18PM -0400, Joseph Adams wrote:
> Update: I'm in the middle of cleaning up the JSON code (
> http://git.postgresql.org/gitweb?p=json-datatype.git;a=summary if you
> want to see the very latest ), so I haven't addressed all of the major
> problems with it yet.
> 
> On Fri, Jul 23, 2010 at 2:34 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> > - I was under the impression that we wanted EXPLAIN (FORMAT JSON) to
> > return type json, but that's obviously not going to be possible if all
> > of this is contrib.  We could (a) move it all into core, (b) move the
> > type itself and its input and output functions into core and leave the
> > rest in contrib [or something along those lines], or (c) give up using
> > it as the return type for EXPLAIN (FORMAT JSON).
> 
> I've been developing it as a contrib module because:
> * I'd imagine it's easier than developing it as a built-in datatype
> right away (e.g. editing a .sql.in file versus editing pg_type.h ).
> * As a module, it has PGXS support, so people can try it out right
> away rather than having to recompile PostgreSQL.
> 
> I, for one, think it would be great if the JSON datatype were all in
> core :-)  However, if and how much JSON code should go into core is up
> for discussion.  Thoughts, anyone?

+1 for putting it in core in 9.1 :)

Cheers,
David.
-- 
David Fetter <david@fetter.org> http://fetter.org/
Phone: +1 415 235 3778  AIM: dfetter666  Yahoo!: dfetter
Skype: davidfetter      XMPP: david.fetter@gmail.com
iCal: webcal://www.tripit.com/feed/ical/people/david74/tripit.ics

Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate

On Wed, 2010-08-11 at 15:27 -0700, David Fetter wrote:
>
> > I've been developing it as a contrib module because:
> > * I'd imagine it's easier than developing it as a built-in datatype
> > right away (e.g. editing a .sql.in file versus editing pg_type.h ).
> > * As a module, it has PGXS support, so people can try it out right
> > away rather than having to recompile PostgreSQL.
> >
> > I, for one, think it would be great if the JSON datatype were all in
> > core :-)  However, if and how much JSON code should go into core is up
> > for discussion.  Thoughts, anyone?
>
> +1 for putting it in core in 9.1 :)

I would be curious to the benefit of putting it in core. I have no
problem with the type but in core?

Joshua D. Drake

--
PostgreSQL.org Major Contributor
Command Prompt, Inc: http://www.commandprompt.com/ - 509.416.6579
Consulting, Training, Support, Custom Development, Engineering
http://twitter.com/cmdpromptinc | http://identi.ca/commandprompt

On Wed, Aug 11, 2010 at 03:40:36PM -0700, Joshua D. Drake wrote:
> On Wed, 2010-08-11 at 15:27 -0700, David Fetter wrote:
> >  
> > > I've been developing it as a contrib module because:
> > > * I'd imagine it's easier than developing it as a built-in
> > > datatype right away (e.g. editing a .sql.in file versus editing
> > > pg_type.h ).
> > > * As a module, it has PGXS support, so people can try it out
> > > right away rather than having to recompile PostgreSQL.
> > > 
> > > I, for one, think it would be great if the JSON datatype were
> > > all in core :-)  However, if and how much JSON code should go
> > > into core is up for discussion.  Thoughts, anyone?
> > 
> > +1 for putting it in core in 9.1 :)
> 
> I would be curious to the benefit of putting it in core. I have no
> problem with the type but in core?

If it's not in core, the vast majority of users will not have it
installed, and nothing, in core or otherwise, will be able to count on
it.

As this is really pretty green-field stuff, it's super unlikely to
break extant code. :)

Cheers,
David.
-- 
David Fetter <david@fetter.org> http://fetter.org/
Phone: +1 415 235 3778  AIM: dfetter666  Yahoo!: dfetter
Skype: davidfetter      XMPP: david.fetter@gmail.com
iCal: webcal://www.tripit.com/feed/ical/people/david74/tripit.ics

Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate

On 08/11/2010 07:33 PM, David Fetter wrote:
>> I would be curious to the benefit of putting it in core. I have no
>> problem with the type but in core?
> If it's not in core, the vast majority of users will not have it
> installed, and nothing, in core or otherwise, will be able to count on
> it.
>
>


You could say that about almost any feature. PostgreSQL is designed to 
be modular, and we can hardly credibly use that as an argument against 
ourselves.

A convincing argument would be that there is another feature we want in 
core that needs or at least could benefit from it.

cheers

andrew

Andrew Dunstan wrote:
> 
> 
> On 08/11/2010 07:33 PM, David Fetter wrote:
> >> I would be curious to the benefit of putting it in core. I have no
> >> problem with the type but in core?
> > If it's not in core, the vast majority of users will not have it
> > installed, and nothing, in core or otherwise, will be able to count on
> > it.
> >
> >
> 
> 
> You could say that about almost any feature. PostgreSQL is designed to 
> be modular, and we can hardly credibly use that as an argument against 
> ourselves.
> 
> A convincing argument would be that there is another feature we want in 
> core that needs or at least could benefit from it.

I would say that JSON is no longer a niche data format, which would
suggest its inclusion in core.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + It's impossible for everything to be true. +

On Wed, Aug 11, 2010 at 07:39:37PM -0400, Andrew Dunstan wrote:
> On 08/11/2010 07:33 PM, David Fetter wrote:
> >>I would be curious to the benefit of putting it in core. I have no
> >>problem with the type but in core?
> >If it's not in core, the vast majority of users will not have it
> >installed, and nothing, in core or otherwise, will be able to count on
> >it.
> 
> You could say that about almost any feature. PostgreSQL is designed
> to be modular, and we can hardly credibly use that as an argument
> against ourselves.
> 
> A convincing argument would be that there is another feature we want
> in core that needs or at least could benefit from it.

EXPLAIN (FORMAT JSON) would benefit right away, as its overall code
would be much more likely to be (and stay) correct.

Cheers,
David.
-- 
David Fetter <david@fetter.org> http://fetter.org/
Phone: +1 415 235 3778  AIM: dfetter666  Yahoo!: dfetter
Skype: davidfetter      XMPP: david.fetter@gmail.com
iCal: webcal://www.tripit.com/feed/ical/people/david74/tripit.ics

Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate

On Wed, 2010-08-11 at 15:27 -0700, David Fetter wrote:
>  
> > I've been developing it as a contrib module because:
> > * I'd imagine it's easier than developing it as a built-in datatype
> > right away (e.g. editing a .sql.in file versus editing pg_type.h ).
> > * As a module, it has PGXS support, so people can try it out right
> > away rather than having to recompile PostgreSQL.
> > 
> > I, for one, think it would be great if the JSON datatype were all in
> > core :-)  However, if and how much JSON code should go into core is up
> > for discussion.  Thoughts, anyone?
> 
> +1 for putting it in core in 9.1 :)

I would be curious to the benefit of putting it in core. I have no
problem with the type but in core?

Joshua D. Drake

-- 
PostgreSQL.org Major Contributor
Command Prompt, Inc: http://www.commandprompt.com/ - 509.416.6579
Consulting, Training, Support, Custom Development, Engineering
http://twitter.com/cmdpromptinc | http://identi.ca/commandprompt

Updated patch:  the JSON code has all been moved into core, so this
patch is now for a built-in data type.  However, I factored the
general-purpose utility functions out into a different patch to be
reviewed separately, so this JSON data type patch won't work without
the utility patch at
http://archives.postgresql.org/pgsql-hackers/2010-08/msg00949.php .

I still need to improve the documentation, as it still looks like the
documentation for a module instead of a built-in data type.  Also, the
code is not taking advantage of the json_type_t enum being built-in
yet; it still uses getEnumLabelOids (one of the utility functions) to
retrieve the OIDs dynamically.


Joey Adams

Hi, I start to review JSON patch.

On Fri, Aug 13, 2010 at 7:33 PM, Joseph Adams
<joeyadams3.14159@gmail.com> wrote:
> Updated patch:  the JSON code has all been moved into core, so this
> patch is now for a built-in data type.

I think the patch can be split into two pieces: 1. Basic I/O support for JSON type (in/out/validate) 2. JSONPath
supportand functions for partial node management 

It is better to submit only 1 at first. Of course we should consider
about JSONPath before deciding the internal representation of JSON,
but separated patches can be easily reviewed.

I have several questions about the spec and implementation.
Sorry if you have already discussed about some of them, but I cannot
understand why the current code is the best design from the patch...

* Should we accept a scalar value as a valid JSON?
According to RFC, the root element of JSON text must be an object
or array. But to_json() and from_json() accept scalar values.

* JSON to a scalar value by from_json()
How about to have json_to_array(JSON) instead of from_json()?
JSON value is always an array or object, it's nonsense to convert
it to a scalar value directly; to an array seems to match better.
In addition, an array can be indexed with GIN; index-able JSON
type is very attractive.

* struct JSON seems to be too complex for me.
Can we use List (pg_list.h) instead of linked-list? 'key' and 'key_length'
fields should be held in the parent's List. i.e, JSON_ARRAY has List of
JSON, and JSON_OBJECT has List of {string, JSON} pairs.

We could also discard 'parent' field. It might be needed by JSONPath,
but we can have parent information in variables on C-stack because we
search JSON trees from root to children, no?

I think we don't need 'orig' field because the original input text is
not so important in normal use cases. Instead, we could have formatter
function something like json_pretty(json) RETURNS text.

* On-disk format of JSON values
(There might be some discussions before... What is the conclusion?)
The current code stores the original input text, but we can use
some kinds of pre-parsed format to store JSON, like hstore.
It can be different from BSON.

* Completeness of JSONPath APIs
json_get() can be replaced with json_path(), no?
Also, we can replace existing nodes with json_set(), but we cannot
append new nodes. What do you think modification of JSON value?
If the design is too difficult, it'd be better only to have search
APIs at this moment. Modification APIs will be added in the future.

--
Itagaki Takahiro

On Wed, Aug 25, 2010 at 1:34 AM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> * Should we accept a scalar value as a valid JSON?
> According to RFC, the root element of JSON text must be an object
> or array. But to_json() and from_json() accept scalar values.

This seems a bit like the XML document/content distinction, which I've
never really understood.  If [[1], false] is a legal JSON value, then
it seems odd that [1] should be legal but false not.

> * JSON to a scalar value by from_json()
> How about to have json_to_array(JSON) instead of from_json()?
> JSON value is always an array or object, it's nonsense to convert
> it to a scalar value directly; to an array seems to match better.
> In addition, an array can be indexed with GIN; index-able JSON
> type is very attractive.

Yeah, I don't like the excessive use of polymorphism either.

> * On-disk format of JSON values
> (There might be some discussions before... What is the conclusion?)
> The current code stores the original input text, but we can use
> some kinds of pre-parsed format to store JSON, like hstore.
> It can be different from BSON.

I see no value to choosing a different on-disk format.  It might speed
up indexing, but I/O will be slower.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

2010/8/25 Robert Haas <robertmhaas@gmail.com>:
> On Wed, Aug 25, 2010 at 1:34 AM, Itagaki Takahiro
> <itagaki.takahiro@gmail.com> wrote:
>> * Should we accept a scalar value as a valid JSON?
>> According to RFC, the root element of JSON text must be an object
>> or array. But to_json() and from_json() accept scalar values.
>
> This seems a bit like the XML document/content distinction, which I've
> never really understood.  If [[1], false] is a legal JSON value, then
> it seems odd that [1] should be legal but false not.

I want false to be parsed without error, just for convinience. JSON
specification seems a bit too strict. For example, it doesn't have
date value as its parts, which results in people implement their own
parsing rule for Date(long). And AFAIK the strictness of JSON parsing
is partly because the main platform was browser engines that can
eval() string that causes security issue. Without execution engine, we
can allow more formats than RFC.

>> * On-disk format of JSON values
>> (There might be some discussions before... What is the conclusion?)
>> The current code stores the original input text, but we can use
>> some kinds of pre-parsed format to store JSON, like hstore.
>> It can be different from BSON.
>
> I see no value to choosing a different on-disk format.  It might speed
> up indexing, but I/O will be slower.

It depends on use cases, but in my mind plain text will do for us. If
we have JavaScript engine in PostgreSQL like pl/v8 and it handles
on-disk format as-is, then we should choose the kind of format, but in
either text or binary format way it is hopeless to have such
compelling environment in the short future.


Regards,

--
Hitoshi Harada

Hitoshi Harada <umi.tanuki@gmail.com> writes:
> It depends on use cases, but in my mind plain text will do for us. If
> we have JavaScript engine in PostgreSQL like pl/v8 and it handles
> on-disk format as-is, then we should choose the kind of format, but in
> either text or binary format way it is hopeless to have such
> compelling environment in the short future.

Well, for javascript support, there's another nice thing happening:
 - plscheme is built on GNU Guile - next version of GNU Guile supports javascript too
 http://plscheme.projects.postgresql.org/ http://wingolog.org/archives/2009/02/22/ecmascript-for-guile

So my current guess at which javascript engine we'd get first would be
plscheme. Now I don't know what implication that would have on the
binary storage format of javascript or json documents.

Regards,
-- 
dim

Hi,

Anyone working on JSON datatype?
If no, I'm going to submit simplified version of JSON datatype patch.


On Wed, Aug 25, 2010 at 2:34 PM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> On Fri, Aug 13, 2010 at 7:33 PM, Joseph Adams
> <joeyadams3.14159@gmail.com> wrote:
>> Updated patch:  the JSON code has all been moved into core, so this
>> patch is now for a built-in data type.
>
> I think the patch can be split into two pieces:
>  1. Basic I/O support for JSON type (in/out/validate)
>  2. JSONPath support and functions for partial node management
>
> It is better to submit only 1 at first. Of course we should consider
> about JSONPath before deciding the internal representation of JSON,
> but separated patches can be easily reviewed.

--
Itagaki Takahiro

On 09/13/2010 09:30 PM, Itagaki Takahiro wrote:
> Hi,
>
> Anyone working on JSON datatype?
> If no, I'm going to submit simplified version of JSON datatype patch.
>

What's the state of the GSOC project?

cheers

andrew

On Tue, Sep 14, 2010 at 04:06, Andrew Dunstan <andrew@dunslane.net> wrote:
>
>
> On 09/13/2010 09:30 PM, Itagaki Takahiro wrote:
>>
>> Hi,
>>
>> Anyone working on JSON datatype?
>> If no, I'm going to submit simplified version of JSON datatype patch.
>>
>
> What's the state of the GSOC project?

Well, GSoC itself is over. But Joey said he'd continue to work on the
patch after that, to get it ready for commit. I've been unable to
confirm that with him for this commitfest though, so I don't know the
exact status.

--
 Magnus Hagander
 Me: http://www.hagander.net/
 Work: http://www.redpill-linpro.com/

On Fri, Aug 13, 2010 at 7:33 PM, Joseph Adams
<joeyadams3.14159@gmail.com> wrote:
> Updated patch:  the JSON code has all been moved into core, so this
> patch is now for a built-in data type.

I have a question about the design of the JSON type. Why do we need to
store the value in UTF8 encoding? It's true the RFC of JSON says the
the encoding SHALL be encoded in Unicode, but I don't understand why
we should reject other encodings.

As I said before, I'd like to propose only 3 features in the commitfest: * TYPE json data type * text to json: FUNCTION
json_parse(text)* json to text: FUNCTION json_stringify(json, whitelist, space) 

JSONPath will be re-implemented on the basic functionalities in the
subsequent commitfest. Do you have a plan to split your patch?
Or, can I continue to develop my patch? If so, JSONPath needs
to be adjusted to the new infrastructure.

I think json_parse() and json_stringify() is well-known APIs for JSON:
https://developer.mozilla.org/En/Using_JSON_in_Firefox
So, it'd be worth buying the names and signatures for our APIs.
(I'll rename json_pretty in my previous patch to json_stringify.)

--
Itagaki Takahiro

On Fri, Sep 17, 2010 at 8:32 AM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> On Fri, Aug 13, 2010 at 7:33 PM, Joseph Adams
> <joeyadams3.14159@gmail.com> wrote:
>> Updated patch:  the JSON code has all been moved into core, so this
>> patch is now for a built-in data type.
>
> I have a question about the design of the JSON type. Why do we need to
> store the value in UTF8 encoding? It's true the RFC of JSON says the
> the encoding SHALL be encoded in Unicode, but I don't understand why
> we should reject other encodings.

Actually, the code in my original patch should work with any server
encoding in PostgreSQL.  However, internally, it operates in UTF-8 and
converts to/from the server encoding when necessary.  I did it this
way because the JSON code needs to handle Unicode escapes like
"\u266B", but there is no simple and efficient way (that I know of) to
convert single characters to/from the server encoding.

I noticed that in your new patch, you sidestepped the encoding issue
by simply storing strings in their encoded form (right?).  This is
nice and simple, but in the future, JSON tree conversions and updates
will still need to deal with the encoding issue somehow.

> As I said before, I'd like to propose only 3 features in the commitfest:
>  * TYPE json data type
>  * text to json: FUNCTION json_parse(text)
>  * json to text: FUNCTION json_stringify(json, whitelist, space)

Although casting from JSON to TEXT does "stringify" it in my original
patch, I think json_stringify would be much more useful.  In addition
to the formatting options, if the internal format of the JSON type
changes and no longer preserves original formatting, then the behavior
of the following would change:

$$    "unnecessary\u0020escape" $$ :: JSON :: TEXT

json_stringify would be more predictable because it would re-encode
the whitespace (but not the \u0020, unless we went out of our way to
make it do that).

Also, json_parse is "unnecessary" if you allow casting from TEXT to
JSON (which my patch does), but I think having json_parse would be
more intuitive for the same reason you do.

Long story short: I like it :-)  If you're keeping track, features
from my patch not in the new code yet are:* Programmatically validating JSON ( json_validate() )* Getting the type of a
JSONvalue ( json_type() )* Converting scalar values to/from JSON* Converting arrays to JSON* JSONPath
 


> JSONPath will be re-implemented on the basic functionalities in the
> subsequent commitfest. Do you have a plan to split your patch?
> Or, can I continue to develop my patch? If so, JSONPath needs
> to be adjusted to the new infrastructure.

I think your patch is on a better footing than mine, so maybe I should
start contributing to your code rather than the other way around.
Before the next commitfest, I could merge the testcases from my patch
in and identify parsing discrepancies (if any).  Afterward, I could
help merge the other features into the new JSON infrastructure.

I can't compile your initial patch against the latest checkout because
json_parser.h and json_scanner.h are missing.  Is there a more recent
patch, or could you update the patch so it compiles?   I'd like to
start tinkering with the new code.  Thanks!


Joey Adams

On Sat, Sep 18, 2010 at 6:45 AM, Joseph Adams
<joeyadams3.14159@gmail.com> wrote:
>>  Why do we need to store the value in UTF8 encoding?
>
>  because the JSON code needs to handle Unicode escapes like
> "\u266B", but there is no simple and efficient way (that I know of) to
> convert single characters to/from the server encoding.

Ah, we don't need UTF8 encoding only to store JSON data, but we should
care about Unicode escape when we support comparison and extracting
values from JSON, right? I see the worth encoding to UTF8.

One of my proposal is we don't have to keep the original input text.
We store JSON data in effective internal formats. If users want to get
human-readable output, they can use stringify() with indentation option.

> I think your patch is on a better footing than mine, so maybe I should
> start contributing to your code rather than the other way around.
> Before the next commitfest, I could merge the testcases from my patch
> in and identify parsing discrepancies (if any).  Afterward, I could
> help merge the other features into the new JSON infrastructure.

Thanks! I'll contribute my codes developed for another project
(PL/JavaScript), and let's merge our codes to the core.

> I can't compile your initial patch against the latest checkout because
> json_parser.h and json_scanner.h are missing.

Hmm, those files should be generated from .y and .l files. I'll check it.

--
Itagaki Takahiro

Itagaki Takahiro wrote:
> On Sat, Sep 18, 2010 at 6:45 AM, Joseph Adams
> <joeyadams3.14159@gmail.com> wrote:
> >>  Why do we need to store the value in UTF8 encoding?
> >
> >  because the JSON code needs to handle Unicode escapes like
> > "\u266B", but there is no simple and efficient way (that I know of) to
> > convert single characters to/from the server encoding.
> 
> Ah, we don't need UTF8 encoding only to store JSON data, but we should
> care about Unicode escape when we support comparison and extracting
> values from JSON, right? I see the worth encoding to UTF8.
> 
> One of my proposal is we don't have to keep the original input text.
> We store JSON data in effective internal formats. If users want to get
> human-readable output, they can use stringify() with indentation option.
> 
> > I think your patch is on a better footing than mine, so maybe I should
> > start contributing to your code rather than the other way around.
> > Before the next commitfest, I could merge the testcases from my patch
> > in and identify parsing discrepancies (if any). ?Afterward, I could
> > help merge the other features into the new JSON infrastructure.
> 
> Thanks! I'll contribute my codes developed for another project
> (PL/JavaScript), and let's merge our codes to the core.
> 
> > I can't compile your initial patch against the latest checkout because
> > json_parser.h and json_scanner.h are missing.
> 
> Hmm, those files should be generated from .y and .l files. I'll check it.

I am please the two efforts can be joined.  I like the idea of
PL/JavaScript too.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + It's impossible for everything to be true. +

On Fri, Sep 17, 2010 at 10:28 PM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> One of my proposal is we don't have to keep the original input text.
> We store JSON data in effective internal formats. If users want to get
> human-readable output, they can use stringify() with indentation option.

There's a trade-off here: this will make some things faster, but other
things slower.  Probably some discussion of the pros and cons is in
order.

(Also, it's important not to break EXPLAIN (FORMAT JSON), which thinks
that the internal format of JSON is text.)

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

On Sat, Sep 18, 2010 at 11:46 AM, Robert Haas <robertmhaas@gmail.com> wrote:
> <itagaki.takahiro@gmail.com> wrote:
>> One of my proposal is we don't have to keep the original input text.
>> We store JSON data in effective internal formats. If users want to get
>> human-readable output, they can use stringify() with indentation option.
>
> There's a trade-off here: this will make some things faster, but other
> things slower.  Probably some discussion of the pros and cons is in
> order.

I didn't intended to introduce non-text internal formats. The original
patch  spent some codes to keep all of whitespaces as-is in the input.
But I'd say we can simplify it.

Except whitespaces, normalization of strings and numbers might be
problem when we support JSON comparison operators -- comparison of
Unicode escaped characters in strings or 0 vs. 0.0 in numbers.

--
Itagaki Takahiro

On Fri, Sep 17, 2010 at 11:12 PM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> On Sat, Sep 18, 2010 at 11:46 AM, Robert Haas <robertmhaas@gmail.com> wrote:
>> <itagaki.takahiro@gmail.com> wrote:
>>> One of my proposal is we don't have to keep the original input text.
>>> We store JSON data in effective internal formats. If users want to get
>>> human-readable output, they can use stringify() with indentation option.
>>
>> There's a trade-off here: this will make some things faster, but other
>> things slower.  Probably some discussion of the pros and cons is in
>> order.
>
> I didn't intended to introduce non-text internal formats. The original
> patch  spent some codes to keep all of whitespaces as-is in the input.
> But I'd say we can simplify it.
>
> Except whitespaces, normalization of strings and numbers might be
> problem when we support JSON comparison operators -- comparison of
> Unicode escaped characters in strings or 0 vs. 0.0 in numbers.

Hmm, yeah.  I'd be tempted to try to keep the user's original
whitespace as far as possible, but disregard it as far as equality
comparison goes.  However, I'm not quite sure what the right thing to
do about 0 vs 0.0 is.  Does the JSON spec say anything about that?

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

On Sat, Sep 18, 2010 at 4:03 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> Hmm, yeah.  I'd be tempted to try to keep the user's original
> whitespace as far as possible, but disregard it as far as equality
> comparison goes.  However, I'm not quite sure what the right thing to
> do about 0 vs 0.0 is.  Does the JSON spec say anything about that?

I didn't find anything in the JSON spec about comparison, but in
JavaScript, 0 == 0.0 and 0 === 0.0 are both true.  Also, JavaScript
considers two arrays or objects equal if and only if they are
references to the same object, meaning [1,2,3] == [1,2,3] is false,
but if you let var a = [1,2,3]; var b = a; , then a == b and a === b
are both true.  Hence, JavaScript can help us when deciding how to
compare scalars, but as for arrays and objects, "we're on our own"
(actually, JSON arrays could be compared lexically like PostgreSQL
arrays already are; I don't think anyone would disagree with that).

I cast my vote for 0 == 0.0 being true.

As for whitespace preservation, I don't think we should go out of our
way to keep it intact.  Sure, preserving formatting for input and
output makes some sense because we'd have to go out of our way to
normalize it, but preserving whitespace in JSONPath tree selections
(json_get) and updates (json_set) is a lot of work (that I shouldn't
have done), and it doesn't really help anybody.  Consider json_get on
a node under 5 levels of indentation.

Another thing to think about is the possibility of using a non-text
format in the future (such as a binary format or even a format that is
internally indexed).  A binary format would likely be faster to
compare (and hence faster to index).  If the JSON data type preserves
whitespace today, it might break assumptions of future code when it
stops preserving whitespace.  This should at least be documented.

On Sat, Sep 18, 2010 at 11:48 PM, Joseph Adams
<joeyadams3.14159@gmail.com> wrote:
> As for whitespace preservation, I don't think we should go out of our
> way to keep it intact.  Sure, preserving formatting for input and
> output makes some sense because we'd have to go out of our way to
> normalize it, but preserving whitespace in JSONPath tree selections
> (json_get) and updates (json_set) is a lot of work (that I shouldn't
> have done), and it doesn't really help anybody.  Consider json_get on
> a node under 5 levels of indentation.

That seems reasonable to me.  I don't mind messing up the whitespace
when someone pulls a value out using a jsonpath, but I don't think we
should mess with it when they just ask us to store a value.  Users
will tend to format their JSON in a way that they find readable, and
while you'd need artificial intelligence to preserve their preferred
formatting in all cases, changing it to what we think is best when
there's no intrinsic necessity doesn't seem helpful.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

On 08/12/2010 06:27 AM, David Fetter wrote:

> +1 for putting it in core in 9.1 :)

There are times I really wish I could get object graphs, or at least 
hierarchically nested object trees, of objects matching various 
criteria. JSON might be a reasonable representation, and one that's 
increasingly well supported by many different clients. Having it core 
would be really handy for that sort of use, especially as managing 
contrib modules is rather far from ideal in Pg as things stand.

Using the usual generic business app example, it'd be good to be able to 
retrieve "customer 1234, as well as all dependent records in Invoices 
and Addresses" with one query, one round trip ... and no horrid ORM-like 
use of left outer joins plus post-processing.

I've been able to do that with nested XML representations, but it's a 
remarkably verbose, bulky way to move chunks of data around.

This patch already looks like it has lots of promise for that sort of 
use. It'd need aggregates, but that's already come up. A 
composite-or-row-type to json converter seems to be much of the point of 
this patch, and that's the only other part that's really required. So 
I'm excited, and I suspect I won't be the only one.

I'm grabbing it to start playing with it now. I just wanted to chime in 
with interest + enthusiasm for JSON as an increasingly useful 
representation.

--
Craig Ringer

2010/10/17 Terry Laurenzo <tj@laurenzo.org>:
> Hi all -
> I independently started some work on a similar capability as was contributed
> back in August by Joey Adams for a json datatype.  Before starting, I did a
> quick search but for some reason didn't turn this existing thread up.
> What I've been working on is out on github for
> now: http://github.com/tlaurenzo/pgjson
> When I started, I was actually aiming for something else, and got caught up
> going down this rabbit hole.  I took a different design approach, making the
> internal form be an extended BSON stream and implementing event-driven
> parsing and serializing to the different formats.  There was some discussion
> in the original thread around storing plain text vs a custom format.  I have
> to admit I've been back and forth a couple of times on this and have come to
> like a BSON-like format for the data at rest.

Reading your proposal, I'm now +1 for BSON-like style. Especially JS
engine's capabilities to map external data to the language
representation are good news. I agree the mapping is engine's task,
not data format task. I'm not sure if your BSON-like format is more
efficient in terms of space and time than plain text, though.

I like as simple design as we can accept. ISTM format, I/O interface,
simple get/set, mapping tuple from/to object, and indexing are minimum
requirement. Something like JSONPath, aggregates, hstore conversion
and whatsoever sound too much.

Regards,

--
Hitoshi Harada

On Sun, Oct 17, 2010 at 5:18 AM, Hitoshi Harada <umi.tanuki@gmail.com> wrote:
> Reading your proposal, I'm now +1 for BSON-like style. Especially JS
> engine's capabilities to map external data to the language
> representation are good news.

Hmm, we could store postgres' data types as-is with their type oids.
I'm not sure whether it is efficient or worth doing, though.

> I like as simple design as we can accept. ISTM format, I/O interface,
> simple get/set, mapping tuple from/to object, and indexing are minimum
> requirement.

+1 to small start, but simple get/set are already debatable...
For example, text/json conversion: A. SELECT '<json>'::json; B. SELECT '<text>'::text::json;

In the git repo, A calls parse_json_to_bson_as_vardata(), so the input
should be a json format. OTOH, B calls pgjson_json_from_text(), so the
input can be any text. Those behaviors are surprising. I think we have
no other choice but to define text-to-json cast as parsing. The same
can be said for json-to-text -- type-output function vs. extracting
text value from json.

I think casting text to/from json should behave in the same way as type
input/output. The xml type works in the same manner. And if so, we might
not have any casts to/from json for consistency, even though there are
no problems in casts for non-text types.


I'll list issues before we start json types even in the simplest cases:
----
1. where to implement json core: external library vs. inner postgres
2. internal format: text vs. binary (*)
3. encoding: always UTF-8 vs. database encoding (*)
4. meaning of casts text to/from json: parse/stringify vs. get/set
5. parser implementation: flex/bison vs. hand-coded.
----
(*) Note that we would have comparison two json values in the future. So,
we might need to normalize the internal format even in text representation.

The most interesting parts of json types, including indexing and jsonpath,
would be made on the json core. We need conclusions about those issues.

-- 
Itagaki Takahiro

On Sat, Oct 16, 2010 at 12:59 PM, Terry Laurenzo <tj@laurenzo.org> wrote:
>    - It is directly iterable without parsing and/or constructing an AST
>    - It is its own representation.  If iterating and you want to tear-off a
> value to be returned or used elsewhere, its a simple buffer copy plus some
> bit twiddling.
>    - It is conceivable that clients already know how to deal with BSON,
> allowing them to work with the internal form directly (ala MongoDB)
>    - It stores a wider range of primitive types than JSON-text.  The most
> important are Date and binary.

When last I looked at that, it appeared to me that what BSON could
represent was a subset of what JSON could represent - in particular,
that it had things like a 32-bit limit on integers, or something along
those lines.  Sounds like it may be neither a superset nor a subset,
in which case I think it's a poor choice for an internal
representation of JSON.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On 10/19/2010 10:44 AM, Robert Haas wrote:
> On Sat, Oct 16, 2010 at 12:59 PM, Terry Laurenzo<tj@laurenzo.org>  wrote:
>>     - It is directly iterable without parsing and/or constructing an AST
>>     - It is its own representation.  If iterating and you want to tear-off a
>> value to be returned or used elsewhere, its a simple buffer copy plus some
>> bit twiddling.
>>     - It is conceivable that clients already know how to deal with BSON,
>> allowing them to work with the internal form directly (ala MongoDB)
>>     - It stores a wider range of primitive types than JSON-text.  The most
>> important are Date and binary.
> When last I looked at that, it appeared to me that what BSON could
> represent was a subset of what JSON could represent - in particular,
> that it had things like a 32-bit limit on integers, or something along
> those lines.  Sounds like it may be neither a superset nor a subset,
> in which case I think it's a poor choice for an internal
> representation of JSON.

Yeah, if it can't handle arbitrary precision numbers as has previously 
been stated it's dead in the water for our purposes, I think.

cheers

andrew

On Tue, Oct 19, 2010 at 11:22 AM, Terry Laurenzo <tj@laurenzo.org> wrote:
> Agreed.  BSON was born out of implementations that either lacked arbitrary
> precision numbers or had a strong affinity to an int/floating point way of
> thinking about numbers.  I believe that if BSON had an arbitrary precision
> number type, it would be a proper superset of JSON.
>
> As an aside, the max range of an int in BSON 64bits.  Back to my original
> comment that BSON was "grown" instead of designed, it looks like both the
> 32bit and 64bit integers were added late in the game and that the original
> designers perhaps were just going to store all numbers as double.
>
> Perhaps we should enumerate the attributes of what would make a good binary
> encoding?

I think we should take a few steps back and ask why we think that
binary encoding is the way to go.  We store XML as text, for example,
and I can't remember any complaints about that on -bugs or
-performance, so why do we think JSON will be different?  Binary
encoding is a trade-off.  A well-designed binary encoding should make
it quicker to extract a small chunk of a large JSON object and return
it; however, it will also make it slower to return the whole object
(because you're adding serialization overhead).  I haven't seen any
analysis of which of those use cases is more important and why.

I am also wondering how this proposed binary encoding scheme will
interact with TOAST.  If the datum is compressed on disk, you'll have
to decompress it anyway to do anything with it; at that point, is
there still going to be a noticeable speed-up from using the binary
encoding?

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On Oct 19, 2010, at 12:17 PM, Robert Haas wrote:

> I think we should take a few steps back and ask why we think that
> binary encoding is the way to go.  We store XML as text, for example,
> and I can't remember any complaints about that on -bugs or
> -performance, so why do we think JSON will be different?  Binary
> encoding is a trade-off.  A well-designed binary encoding should make
> it quicker to extract a small chunk of a large JSON object and return
> it; however, it will also make it slower to return the whole object
> (because you're adding serialization overhead).  I haven't seen any
> analysis of which of those use cases is more important and why.

Maybe someone has numbers on that for the XML type?

Best,

David

On Tue, Oct 19, 2010 at 11:22 AM, Terry Laurenzo <tj@laurenzo.org> wrote:
> Perhaps we should enumerate the attributes of what would make a good binary
> encoding?

Not sure if we're discussing the internal storage format or the binary
send/recv format, but in my humble opinion, some attributes of a good
internal format are:

1. Lightweight - it'd be really nice for the JSON datatype to be
available in core (even if extra features like JSONPath aren't).
2. Efficiency - Retrieval and storage of JSON datums should be
efficient.  The internal format should probably closely resemble the
binary send/recv format so there's a good reason to use it.

A good attribute of the binary send/recv format would be
compatibility.  For instance, if MongoDB (which I know very little
about) has binary send/receive, perhaps the JSON data type's binary
send/receive should use it.

Efficient retrieval and update of values in a large JSON tree would be
cool, but would be rather complex, and IMHO, overkill.  JSON's main
advantage is that it's sort of a least common denominator of the type
systems of many popular languages, making it easy to transfer
information between them.  Having hierarchical key/value store support
would be pretty cool, but I don't think it's what PostgreSQL's JSON
data type should do.

On Tue, Oct 19, 2010 at 3:17 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> I think we should take a few steps back and ask why we think that
> binary encoding is the way to go.  We store XML as text, for example,
> and I can't remember any complaints about that on -bugs or
> -performance, so why do we think JSON will be different?  Binary
> encoding is a trade-off.  A well-designed binary encoding should make
> it quicker to extract a small chunk of a large JSON object and return
> it; however, it will also make it slower to return the whole object
> (because you're adding serialization overhead).  I haven't seen any
> analysis of which of those use cases is more important and why.

Speculation: the overhead involved with retrieving/sending and
receiving/storing JSON (not to mention TOAST
compression/decompression) will be far greater than that of
serializing/unserializing.

On Tue, Oct 19, 2010 at 3:40 PM, Joseph Adams
<joeyadams3.14159@gmail.com> wrote:
> On Tue, Oct 19, 2010 at 3:17 PM, Robert Haas <robertmhaas@gmail.com> wrote:
>> I think we should take a few steps back and ask why we think that
>> binary encoding is the way to go.  We store XML as text, for example,
>> and I can't remember any complaints about that on -bugs or
>> -performance, so why do we think JSON will be different?  Binary
>> encoding is a trade-off.  A well-designed binary encoding should make
>> it quicker to extract a small chunk of a large JSON object and return
>> it; however, it will also make it slower to return the whole object
>> (because you're adding serialization overhead).  I haven't seen any
>> analysis of which of those use cases is more important and why.
>
> Speculation: the overhead involved with retrieving/sending and
> receiving/storing JSON (not to mention TOAST
> compression/decompression) will be far greater than that of
> serializing/unserializing.

I speculate that your speculation is incorrect.  AIUI, we, unlike
$COMPETITOR, tend to be CPU-bound rather than IO-bound on COPY.  But
perhaps less speculation and more benchmarking is in order.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On Tue, Oct 19, 2010 at 12:17 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> I think we should take a few steps back and ask why we think that
> binary encoding is the way to go.  We store XML as text, for example,
> and I can't remember any complaints about that on -bugs or
> -performance, so why do we think JSON will be different?  Binary
> encoding is a trade-off.  A well-designed binary encoding should make
> it quicker to extract a small chunk of a large JSON object and return
> it; however, it will also make it slower to return the whole object
> (because you're adding serialization overhead).  I haven't seen any
> analysis of which of those use cases is more important and why.
>

The elephant in the room is if the binary encoded form is smaller then
it occupies less ram and disk bandwidth to copy it around. If your
database is large that alone is the dominant factor. Doubling the size
of all the objects in your database means halving the portion of the
database that fits in RAM and doubling the amount of I/O required to
complete any given operation. If your database fits entirely in RAM
either way then it still means less RAM bandwidth used which is often
the limiting factor but depending on how much cpu effort it takes to
serialize and deserialize the balance could shift either way.




--
greg

>
> After spending a week in the morass of this, I have to say that I am less
> certain than I was on any front regarding the text/binary distinction.  I'll
> take some time and benchmark different cases.  My hypothesis is that a well
> implemented binary structure and conversions will add minimal overhead in
> the IO + Validate case which would be the typical in/out flow.  It could be
> substantially faster for binary send/receive because the validation step
> could be eliminated/reduced.  Further storing as binary reduces the overhead
> of random access to the data by database functions.
>
> I'm envisioning staging this up as follows:
>    1. Create a "jsontext".  jsontext uses text as its internal
> representation.  in/out functions are essentially a straight copy or a copy
> + validate.
>    2. Create a "jsonbinary" type.  This uses an optimized binary format for
> internal rep and send/receive.  in/out is a parse/transcode operation to
> standard JSON text.
>    3. Internal data access functions and JSON Path require a jsonbinary.
>    4. There are implicit casts to/from jsontext and jsonbinary.

some years ago I solved similar problems with xml type. I think, so
you have to calculate with two factors:

a) all varlena types are compressed - you cannot to get some
interesting fragment or you cannot tu update some interesting
fragment, every time pg working with complete document

b) access to some fragment of JSON or XML document are not really
important, because fast access to data are solved via indexes.

c) only a few API allows binary communication between server/client.
Almost all interfases use only text based API. I see some possible
interesting direction for binary protocol when some one uses a
javascript driver, when some one use pg in some javascript server side
environment, but it isn't a often used now.

Regards

Pavel

>
> I've got a grammar in mind for the binary structure that I'll share later
> when I've got some more time.  It's inspired by $COMPETITOR's format but a
> little more sane, using type tags that implicitly define the size of the
> operands, simplifying parsing.
>
> I'll then define the various use cases and benchmark using the different
> types.  Some examples include such as IO No Validate, IO+Validate, Store and
> Index, Internal Processing, Internal Composition, etc.
>
> The answer may be to have both a jsontext and jsonbinary type as each will
> be optimized for a different case.
>
> Make sense?  It may be a week before I get through this.
> Terry
>
>

2010/10/19 Greg Stark <gsstark@mit.edu>:
> On Tue, Oct 19, 2010 at 12:17 PM, Robert Haas <robertmhaas@gmail.com> wrote:
>> I think we should take a few steps back and ask why we think that
>> binary encoding is the way to go.  We store XML as text, for example,
>> and I can't remember any complaints about that on -bugs or
>> -performance, so why do we think JSON will be different?  Binary
>> encoding is a trade-off.  A well-designed binary encoding should make
>> it quicker to extract a small chunk of a large JSON object and return
>> it; however, it will also make it slower to return the whole object
>> (because you're adding serialization overhead).  I haven't seen any
>> analysis of which of those use cases is more important and why.
>>
>
> The elephant in the room is if the binary encoded form is smaller then
> it occupies less ram and disk bandwidth to copy it around. If your
> database is large that alone is the dominant factor. Doubling the size
> of all the objects in your database means halving the portion of the
> database that fits in RAM and doubling the amount of I/O required to
> complete any given operation. If your database fits entirely in RAM
> either way then it still means less RAM bandwidth used which is often
> the limiting factor but depending on how much cpu effort it takes to
> serialize and deserialize the balance could shift either way.

I am not sure, if this argument is important for json. This protocol
has not big overhead and json documents are pretty small. More - from
9.0 TOAST uses a relative aggresive compression. I would to like a
some standardised format for json inside pg too, but without using a
some external library I don't see a advantages to use a other format
then text.

Regards

Pavel

>
>
>
>
> --
> greg
>
> --
> Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-hackers
>

Terry Laurenzo <tj@laurenzo.org> writes:
> After spending a week in the morass of this, I have to say that I am less
> certain than I was on any front regarding the text/binary distinction.  I'll
> take some time and benchmark different cases.  My hypothesis is that a well
> implemented binary structure and conversions will add minimal overhead in
> the IO + Validate case which would be the typical in/out flow.  It could be
> substantially faster for binary send/receive because the validation step
> could be eliminated/reduced.

I think that arguments proceeding from speed of binary send/receive
aren't worth the electrons they're written on, because there is nothing
anywhere that says what the binary format ought to be.  In the case of
XML we're just using the text representation as the binary format too,
and nobody's complained about that.  If we were to choose to stick with
straight text internally for a JSON type, we'd do the same thing, and
again nobody would complain.

So, if you want to make a case for using some binary internal format or
other, make it without that consideration.

> I'm envisioning staging this up as follows:
>    1. Create a "jsontext".  jsontext uses text as its internal
> representation.  in/out functions are essentially a straight copy or a copy
> + validate.
>    2. Create a "jsonbinary" type.  This uses an optimized binary format for
> internal rep and send/receive.  in/out is a parse/transcode operation to
> standard JSON text.

Ugh.  Please don't.  JSON should be JSON, and nothing else.  Do you see
any other datatypes in Postgres that expose such internal considerations?
        regards, tom lane

Greg Stark <gsstark@mit.edu> writes:
> The elephant in the room is if the binary encoded form is smaller then
> it occupies less ram and disk bandwidth to copy it around.

It seems equally likely that a binary-encoded form could be larger
than the text form (that's often true for our other datatypes).
Again, this is an argument that would require experimental evidence
to back it up.
        regards, tom lane

<div class="gmail_quote">On Tue, Oct 19, 2010 at 4:51 PM, Tom Lane <span dir="ltr"><<a
href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>></span>wrote:<br /><blockquote class="gmail_quote"
style="border-left:1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div class="im">Terry
Laurenzo<<a href="mailto:tj@laurenzo.org">tj@laurenzo.org</a>> writes:<br /> > After spending a week in the
morassof this, I have to say that I am less<br /> > certain than I was on any front regarding the text/binary
distinction. I'll<br /> > take some time and benchmark different cases.  My hypothesis is that a well<br /> >
implementedbinary structure and conversions will add minimal overhead in<br /> > the IO + Validate case which would
bethe typical in/out flow.  It could be<br /> > substantially faster for binary send/receive because the validation
step<br/> > could be eliminated/reduced.<br /><br /></div>I think that arguments proceeding from speed of binary
send/receive<br/> aren't worth the electrons they're written on, because there is nothing<br /> anywhere that says what
thebinary format ought to be.  In the case of<br /> XML we're just using the text representation as the binary format
too,<br/> and nobody's complained about that.  If we were to choose to stick with<br /> straight text internally for a
JSONtype, we'd do the same thing, and<br /> again nobody would complain.<br /><br /> So, if you want to make a case for
usingsome binary internal format or<br /> other, make it without that consideration.<br /><div class="im"><br /> >
I'menvisioning staging this up as follows:<br /> >    1. Create a "jsontext".  jsontext uses text as its internal<br
/>> representation.  in/out functions are essentially a straight copy or a copy<br /> > + validate.<br /> >  
 2.Create a "jsonbinary" type.  This uses an optimized binary format for<br /> > internal rep and send/receive.
 in/outis a parse/transcode operation to<br /> > standard JSON text.<br /><br /></div>Ugh.  Please don't.  JSON
shouldbe JSON, and nothing else.  Do you see<br /> any other datatypes in Postgres that expose such internal
considerations?<br/><br />                        regards, tom lane<br /></blockquote></div><br />I don't think anyone
herewas really presenting arguments as yet.  We're several layers deep on speculation and everyone is saying that
experimentationis needed.<br /><br />I've got my own reasons for going down this path for a solution I have in mind.  I
hadthought that some part of that might have been applicable to pg core, but if not, that's no problem.  For my own
edification,I'm going to proceed down this path and see where it leads.  I'll let the list know what I find out.<br
/><br/>I can understand the sentiment that JSON should be JSON and nothing else from a traditional database server's
pointof view, but there is nothing sacrosanct about it in the broader context.<br /><br />Terry<br /><br /> 

Excerpts from David E. Wheeler's message of mar oct 19 16:36:20 -0300 2010:
> On Oct 19, 2010, at 12:17 PM, Robert Haas wrote:
> 
> > I think we should take a few steps back and ask why we think that
> > binary encoding is the way to go.  We store XML as text, for example,
> > and I can't remember any complaints about that on -bugs or
> > -performance, so why do we think JSON will be different?  Binary
> > encoding is a trade-off.  A well-designed binary encoding should make
> > it quicker to extract a small chunk of a large JSON object and return
> > it; however, it will also make it slower to return the whole object
> > (because you're adding serialization overhead).  I haven't seen any
> > analysis of which of those use cases is more important and why.
> 
> Maybe someone has numbers on that for the XML type?

Like these?
http://exificient.sourceforge.net/?id=performance

-- 
Álvaro Herrera <alvherre@commandprompt.com>
The PostgreSQL Company - Command Prompt, Inc.
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

On Tue, Oct 19, 2010 at 6:56 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Greg Stark <gsstark@mit.edu> writes:
>> The elephant in the room is if the binary encoded form is smaller then
>> it occupies less ram and disk bandwidth to copy it around.
>
> It seems equally likely that a binary-encoded form could be larger
> than the text form (that's often true for our other datatypes).
> Again, this is an argument that would require experimental evidence
> to back it up.

That's exactly what I was thinking when I read Greg's email.  I
designed something vaguely (very vaguely) like this many years ago and
the binary format that I worked so hard to create was enormous
compared to the text format, mostly because I had a lot of small
integers in the data I was serializing, and as it turns out,
representing {0,1,2} in less than 7 bytes is not very easy.  It can
certainly be done if you set out to optimize for precisely those kinds
of cases, but I ended up with something awful like:

<4 byte type = list> <4 byte list length = 3> <4 byte type = integer>
<4 byte integer = 0> <4 byte type = integer> <4 byte integer = 1> <4
byte type = integer> <4 byte integer = 2>

= 32 bytes.  Even if you were a little smarter than I was and used 2
byte integers (with some escape hatch allowing larger numbers to be
represented) it's still more than twice the size of the text
representation.  Even if you use 1 byte integers it's still bigger.
To get it down to being smaller, you've got to do something like make
the high nibble of each byte a type field and the low nibble the first
4 payload bits.  You can certainly do all of this but you could also
just store it as text and let the TOAST compression algorithm worry
about making it smaller.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

I hear ya.  It might be a premature optimization but I still think there may be benefit for the case of large scale
extractionand in- database transformation of large JSON datastructures.  We have terabytes of this stuff and I'd like
somethingbetween the hip nosql options and a fully structured SQL datastore. 

Terry

Sent from my iPhone

On Oct 19, 2010, at 6:36 PM, Robert Haas <robertmhaas@gmail.com> wrote:

> On Tue, Oct 19, 2010 at 6:56 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Greg Stark <gsstark@mit.edu> writes:
>>> The elephant in the room is if the binary encoded form is smaller then
>>> it occupies less ram and disk bandwidth to copy it around.
>>
>> It seems equally likely that a binary-encoded form could be larger
>> than the text form (that's often true for our other datatypes).
>> Again, this is an argument that would require experimental evidence
>> to back it up.
>
> That's exactly what I was thinking when I read Greg's email.  I
> designed something vaguely (very vaguely) like this many years ago and
> the binary format that I worked so hard to create was enormous
> compared to the text format, mostly because I had a lot of small
> integers in the data I was serializing, and as it turns out,
> representing {0,1,2} in less than 7 bytes is not very easy.  It can
> certainly be done if you set out to optimize for precisely those kinds
> of cases, but I ended up with something awful like:
>
> <4 byte type = list> <4 byte list length = 3> <4 byte type = integer>
> <4 byte integer = 0> <4 byte type = integer> <4 byte integer = 1> <4
> byte type = integer> <4 byte integer = 2>
>
> = 32 bytes.  Even if you were a little smarter than I was and used 2
> byte integers (with some escape hatch allowing larger numbers to be
> represented) it's still more than twice the size of the text
> representation.  Even if you use 1 byte integers it's still bigger.
> To get it down to being smaller, you've got to do something like make
> the high nibble of each byte a type field and the low nibble the first
> 4 payload bits.  You can certainly do all of this but you could also
> just store it as text and let the TOAST compression algorithm worry
> about making it smaller.
>
> --
> Robert Haas
> EnterpriseDB: http://www.enterprisedb.com
> The Enterprise PostgreSQL Company

On Tue, Oct 19, 2010 at 9:15 PM, Terri Laurenzo <tj@laurenzo.org> wrote:
> I hear ya.  It might be a premature optimization but I still think there may be benefit for the case of large scale
extractionand in- database transformation of large JSON datastructures.  We have terabytes of this stuff and I'd like
somethingbetween the hip nosql options and a fully structured SQL datastore. 

Well, keep hacking on it!

I'm interested to hear what you find out.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On Wed, Oct 20, 2010 at 6:39 AM, Terry Laurenzo <tj@laurenzo.org> wrote:
> The answer may be to have both a jsontext and jsonbinary type as each will
> be optimized for a different case.

I want to choose one format for JSON rather than having two types.
It should be more efficient than other format in many cases,
and not so bad in other cases.

I think the discussion was started with "BSON could represent was a subset of what JSON could represent".
So, any binary format could be acceptable that have enough
representational power compared with text format.

For example, a sequence of <byte-length> <text> could reduce
CPU cycles for reparsing and hold all of the input as-is except
ignorable white-spaces. It is not a BSON, but is a binary format.

Or, if we want to store numbers in binary form, I think the
format will be numeric type in postgres. It has high precision,
and we don't need any higher precision than it to compare two
numbers eventually. Even if we use BSON format, we need to extend
it to store all of numeric values, that precision is 10^1000.

-- 
Itagaki Takahiro

Good points.  In addition, any binary format needs to support object property traversal without having to do a deep
scanof all descendants.  BSON handles this with explicit lengths for document types (objects and arrays) so that entire
partsof the tree can be skipped during sibling traversal.<br /><br />It would also be nice to make sure that we store
fullyparsed strings.  There are lots of escape options that simply do not need to be preserved (c escapes, unicode,
octal,hex sequences) and hinder the ability to do direct comparisons.  BSON also makes a small extra effort to ensure
thatobject property names are encoded in a way that is easily comparable, as this will be the most frequently compared
items.<br/><br />I'm still going to write up a proposed grammar that takes these items into account - just ran out of
timetonight.<br /><br />Terry<br /><br /><div class="gmail_quote">On Wed, Oct 20, 2010 at 12:46 AM, Itagaki Takahiro
<spandir="ltr"><<a href="mailto:itagaki.takahiro@gmail.com">itagaki.takahiro@gmail.com</a>></span> wrote:<br
/><blockquoteclass="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex;
padding-left:1ex;"><div class="im">On Wed, Oct 20, 2010 at 6:39 AM, Terry Laurenzo <<a
href="mailto:tj@laurenzo.org">tj@laurenzo.org</a>>wrote:<br /> > The answer may be to have both a jsontext and
jsonbinarytype as each will<br /> > be optimized for a different case.<br /><br /></div>I want to choose one format
forJSON rather than having two types.<br /> It should be more efficient than other format in many cases,<br /> and not
sobad in other cases.<br /><br /> I think the discussion was started with<br />  "BSON could represent was a subset of
whatJSON could represent".<br /> So, any binary format could be acceptable that have enough<br /> representational
powercompared with text format.<br /><br /> For example, a sequence of <byte-length> <text> could reduce<br
/>CPU cycles for reparsing and hold all of the input as-is except<br /> ignorable white-spaces. It is not a BSON, but
isa binary format.<br /><br /> Or, if we want to store numbers in binary form, I think the<br /> format will be numeric
typein postgres. It has high precision,<br /> and we don't need any higher precision than it to compare two<br />
numberseventually. Even if we use BSON format, we need to extend<br /> it to store all of numeric values, that
precisionis 10^1000.<br /><br /> --<br /><font color="#888888">Itagaki Takahiro<br /></font></blockquote></div><br /> 

* Terry Laurenzo:

> Agreed.  BSON was born out of implementations that either lacked
> arbitrary precision numbers or had a strong affinity to an
> int/floating point way of thinking about numbers.  I believe that if
> BSON had an arbitrary precision number type, it would be a proper
> superset of JSON.

But JSON has only double-precision numbers!?

On 10/20/2010 01:15 PM, Florian Weimer wrote:
> * Terry Laurenzo:
>
>> Agreed.  BSON was born out of implementations that either lacked
>> arbitrary precision numbers or had a strong affinity to an
>> int/floating point way of thinking about numbers.  I believe that if
>> BSON had an arbitrary precision number type, it would be a proper
>> superset of JSON.
> But JSON has only double-precision numbers!?

AFAICT the JSON spec says nothing at all about the precision of numbers. 
It just provides a syntax for them. We should not confuse what can be 
allowed in JSON with what can be handled by some consumers of JSON such 
as ECMAScript.

However, since we would quite reasonably require that any JSON 
implementation be able to handle arbitrary precision numbers, that 
apparently rules out BSON as a storage engine for it, since BSON can not 
handle such things.


cheers

andrew

On Sat, Oct 23, 2010 at 8:33 PM, Terry Laurenzo <tj@laurenzo.org> wrote:
>>
>> I'm still going to write up a proposed grammar that takes these items into
>> account - just ran out of time tonight.
>>
>
> The binary format I was thinking of is here:
>
>   http://github.com/tlaurenzo/pgjson/blob/master/pgjson/shared/include/json/jsonbinary.h
> This was just a quick brain dump and I haven't done a lot of diligence on
> verifying it, but I think it should be more compact than most JSON text
> payloads and quick to iterate over/update in sibling traversal order vs
> depth-first traversal which is what we would get out of JSON text.
> Thoughts?
> Terry

It doesn't do particularly well on my previous example of [1,2,3].  It
comes out slightly shorter on ["a","b","c"] and better if the strings
need any escaping.  I don't think the float4 and float8 formats are
very useful; how could you be sure that the output was going to look
the same as the input?  Or alternatively that dumping a particular
object to text and reloading it will produce the same internal
representation?  I think it would be simpler to represent integers
using a string of digits; that way you can be assured of going from
text -> binary -> text without change.

Perhaps it would be enough to define the high two bits as follows: 00
= array, 01 = object, 10 = string, 11 = number/true/false/null.  The
next 2 bits specify how the length is stored.  00 = remaining 4 bits
store a length of up to 15 bytes, 01 = remaining 4 bits + 1 additional
byte store a 12-bit length of up to 4K, 10 = remaining 4 bits + 2
additional bytes store a 20-bit length of up to 1MB, 11 = 4 additional
bytes store a full 32-bit length word.  Then, the array, object, and
string representations can work as you've specified them.  Anything
else can be represented by itself, or perhaps we should say that
numbers represent themselves and true/false/null are represented by a
1-byte sequence, t/f/n (or perhaps we could define 111100{00,01,10} to
mean those values, since there's no obvious reason for the low bits to
be non-zero if a 4-bit length word ostensibly follows).

So [1,2,3] = 06 C1 '1' C1 '2' C1 '3' and ["a","b","c"] = 06 81 'a' 81 'b' 81 'c'

(I am still worried about the serialization/deserialization overhead
but that's a different issue.)

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On Sun, Oct 24, 2010 at 2:21 AM, Terry Laurenzo <tj@laurenzo.org> wrote:
>> It doesn't do particularly well on my previous example of [1,2,3].  It
>> comes out slightly shorter on ["a","b","c"] and better if the strings
>> need any escaping.  I don't think the float4 and float8 formats are
>> very useful; how could you be sure that the output was going to look
>> the same as the input?  Or alternatively that dumping a particular
>> object to text and reloading it will produce the same internal
>> representation?  I think it would be simpler to represent integers
>> using a string of digits; that way you can be assured of going from
>> text -> binary -> text without change.
>>
>> Perhaps it would be enough to define the high two bits as follows: 00
>> = array, 01 = object, 10 = string, 11 = number/true/false/null.  The
>> next 2 bits specify how the length is stored.  00 = remaining 4 bits
>> store a length of up to 15 bytes, 01 = remaining 4 bits + 1 additional
>> byte store a 12-bit length of up to 4K, 10 = remaining 4 bits + 2
>> additional bytes store a 20-bit length of up to 1MB, 11 = 4 additional
>> bytes store a full 32-bit length word.  Then, the array, object, and
>> string representations can work as you've specified them.  Anything
>> else can be represented by itself, or perhaps we should say that
>> numbers represent themselves and true/false/null are represented by a
>> 1-byte sequence, t/f/n (or perhaps we could define 111100{00,01,10} to
>> mean those values, since there's no obvious reason for the low bits to
>> be non-zero if a 4-bit length word ostensibly follows).
>>
>> So [1,2,3] = 06 C1 '1' C1 '2' C1 '3' and ["a","b","c"] = 06 81 'a' 81 'b'
>> 81 'c'
>>
>> (I am still worried about the serialization/deserialization overhead
>> but that's a different issue.)
>>
>
> Thanks.  I'll play around with the bit and numeric encodings you've
> recommended.  Arrays are certainly the toughest to economize on as a text
> encoding has minimum of 2 + n - 1 overhead bytes.  Text encoding for objects
> has some more wiggle room with 2 + (n-1) + (n*3) extra bytes.  I was
> admittedly thinking about more complicated objects.
> I'm still worried about transcoding overhead as well.  If comparing to a
> simple blind storage of JSON text with no validation or normalization, there
> is obviously no way to beat a straight copy.  However, its not outside the
> realm of reason to think that it may be possible to match or beat the clock
> if comparing against text to text normalization, or perhaps adding slight
> overhead to a validator.  The advantage to the binary structure is the
> ability to scan it hierarchically in sibling order and provide mutation
> operations with simple memcpy's as opposed to parse_to_ast -> modify_ast ->
> serialize_ast.

Yeah, my concern is not whether the overhead will be zero; it's
whether it will be small, yet allow large gains on other operations.
Like, how much slower will it be to pull out a moderately complex 1MB
JSON blob (not just a big string) out of a single-row, single-column
table?  If it's 5% slower, that's probably OK, since this is a
reasonable approximation of a worst-case scenario.  If it's 50%
slower, that sounds painful.  It would also be worth testing with a
much smaller size, such as a 1K object with lots of internal
structure.  In both cases, all data cached in shared_buffers, etc.

Then on the flip side how do we do on val[37]["whatever"]?  You'd like
to hope that this will be significantly faster than the text encoding
on both large and small objects.  If it's not, there's probably not
much point.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Are there any activities in JSON data types for the next commitfest?
I'd like to help you if you are  working on the topic,
or I can make up previous works and discussions by myself.

On Mon, Oct 25, 2010 at 1:36 AM, Terry Laurenzo <tj@laurenzo.org> wrote:
>> Yeah, my concern is not whether the overhead will be zero; it's
>> whether it will be small, yet allow large gains on other operations.
>> Like, how much slower will it be to pull out a moderately complex 1MB
>> JSON blob (not just a big string) out of a single-row, single-column
>> table?  If it's 5% slower, that's probably OK, since this is a
>> reasonable approximation of a worst-case scenario.  If it's 50%
>> slower, that sounds painful.  It would also be worth testing with a
>> much smaller size, such as a 1K object with lots of internal
>> structure.  In both cases, all data cached in shared_buffers, etc.
>>
>> Then on the flip side how do we do on val[37]["whatever"]?  You'd like
>> to hope that this will be significantly faster than the text encoding
>> on both large and small objects.  If it's not, there's probably not
>> much point.
>
> We're on the same page.  I'm implementing the basic cases now and then will
> come up with some benchmarks.

--
Itagaki Takahiro

On Mon, Nov 8, 2010 at 9:52 PM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> Are there any activities in JSON data types for the next commitfest?

I'm leaning toward the view that we shouldn't commit a JSON
implementation to core (or contrib) for 9.1.  We have at least three
competing proposals on the table.  I thought of picking it up and
hacking on it myself, but then we'd have four competing proposals on
the table.  Even if we could come to some consensus on which of those
proposals is technically superior, the rate at which new ideas are
being proposed suggests to me that it would be premature to anoint any
single implementation as our canonical one.  I'd like to see some of
these patches finished and put up on pgfoundry or github, and then
consider moving one of them into core when we have a clear and stable
consensus that one of them is technically awesome and the best thing
we're going to get.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On Sun, Nov 21, 2010 at 12:31 AM, Terry Laurenzo <tj@laurenzo.org> wrote:
> What we see from this is that parsing JSON text and generating a binary
> representation is cheap, representing approximately 10% of the base case
> time.  Conversely, anything that involves generating JSON text is expensive,
> accounting for 30-40% of the base case time.  Some incidental profiling
> shows that while the entire operation is expensive, the process of
> generating string literals dominates this time.  There is likely room for
> optimization in this method, but it should be noted that most of these
> documents are lightly escaped (if escaped at all) which represents the happy
> path through the string literal output function.

Ouch!  That's kind of painful.  But certainly for some use cases it
will work out to a huge speedup, if you're doing subscripting or
similar.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On Nov 20, 2010, at 9:31 PM, Terry Laurenzo wrote:

> Assuming that the JSON datatype (at a minimum) normalizes text for storage, then the text storage option accounts for
aboutthe most expensive path but with none of the benefits of an internal binary form (smaller size, ability to cheaply
performnon-trivial manipulation within the database server). 
>
> Of course, just having a JSON datatype that blindly stores text will beat everything, but I'm getting closer to
thinkingthat the binary option is worth the tradeoff. 
>
> Comments?

benchmarks++

Nice to have some data points for this discussion.

Best,

David, still hoping for the JSON data type in 9.1…

I copied the 5 sample documents from json.org's example section for these tests.  These are loaded into a table with a varchar column 1000 times each (so the test table has 5000 rows in it).  In all situations, the binary encoding was smaller than the normalized text form (between 9 and 23% smaller).  I think there are cases where the binary form will be larger than the corresponding text form, but I don't think they would be very common.