From tsufiev at gmail.com  Tue Aug 25 10:12:30 2009
From: tsufiev at gmail.com (Timur Sufiev)
Date: Tue, 25 Aug 2009 14:12:30 +0400
Subject: [postmodern-devel] [S-SQL] Multiple rows in INSERT INTO clause
Message-ID: <4a93b8f8.11355e0a.4ae3.2c89@mx.google.com>

Is there some way in S-SQL to generate INSERT INTO clauses with multiple
rows, e.g. "INSERT INTO \"table\" (a, b) VALUES (10, 20), (30, 40)" ?
Desired sexp format is `(:insert-into table :set a (10 20) b (30 40)),
transformation can work like a MAPCAR: stop processing values sequences
as soon as the end of the shortest one is reached. 

The reason to have such a feature is a performance issue: having a sequence
of rows to be inserted we want to minimize the overhead related with
sending queries to db; one query inserting, say, 50 rows is certainly
faster than 50 queries each inserting 1 row.

I would prepare the patch myself, but the syntax of s-sql.lisp is a bit too
tricky for me :(. 

-- 
Timur Sufiev

-- 
Timur Sufiev



From daniel at whitehouse.id.au  Tue Aug 25 12:11:18 2009
From: daniel at whitehouse.id.au (Daniel White)
Date: Tue, 25 Aug 2009 22:11:18 +1000
Subject: [postmodern-devel] [S-SQL] Multiple rows in INSERT INTO clause
In-Reply-To: <4a93b8f8.11355e0a.4ae3.2c89@mx.google.com>
References: <4a93b8f8.11355e0a.4ae3.2c89@mx.google.com>
Message-ID: <20090825221118.6a436f61@whitehouse.id.au>

On Tue, 25 Aug 2009 14:12:30 +0400
Timur Sufiev <tsufiev at gmail.com> wrote:

> Is there some way in S-SQL to generate INSERT INTO clauses with multiple
> rows, e.g. "INSERT INTO \"table\" (a, b) VALUES (10, 20), (30, 40)" ?
> Desired sexp format is `(:insert-into table :set a (10 20) b (30 40)),

The functionality already exists, albeit undocumented.

(:insert-rows-into 'table
                   :columns 'a 'b
                   :values '((10 20) (30 40)))

-- 
Daniel White



From tsufiev at gmail.com  Tue Aug 25 12:31:18 2009
From: tsufiev at gmail.com (Timur Sufiev)
Date: Tue, 25 Aug 2009 16:31:18 +0400
Subject: [postmodern-devel] [S-SQL] Multiple rows in INSERT INTO clause
In-Reply-To: <20090825221118.6a436f61@whitehouse.id.au>
References: <4a93b8f8.11355e0a.4ae3.2c89@mx.google.com>
	<20090825221118.6a436f61@whitehouse.id.au>
Message-ID: <4a93d981.01145e0a.104e.53d1@mx.google.com>

>> Is there some way in S-SQL to generate INSERT INTO clauses with multiple
>> rows, e.g. "INSERT INTO \"table\" (a, b) VALUES (10, 20), (30, 40)" ?
>> Desired sexp format is `(:insert-into table :set a (10 20) b (30 40)),
> 
> The functionality already exists, albeit undocumented.
> 
> (:insert-rows-into 'table
>                    :columns 'a 'b
>                    :values '((10 20) (30 40)))
> 
> -- 
> Daniel White
Thanks, Daniel!
-- 
Timur Sufiev



From daniel at whitehouse.id.au  Tue Aug 25 12:48:48 2009
From: daniel at whitehouse.id.au (Daniel White)
Date: Tue, 25 Aug 2009 22:48:48 +1000
Subject: [postmodern-devel] Add :add-column action to :alter-table
Message-ID: <20090825224848.666530ef@whitehouse.id.au>

Patch attached adds an :add-column action to :alter-table where the
column should be the same form as a column in :create-table.

For example:

(:alter-table table
              :add-column col-name :type string :default "default"))

-- 
Daniel White
-------------- next part --------------
A non-text attachment was scrubbed...
Name: add-column.patch
Type: text/x-patch
Size: 33397 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/postmodern-devel/attachments/20090825/04639a01/attachment.bin>

From marijnh at gmail.com  Tue Aug 25 16:13:23 2009
From: marijnh at gmail.com (Marijn Haverbeke)
Date: Tue, 25 Aug 2009 18:13:23 +0200
Subject: [postmodern-devel] Add :add-column action to :alter-table
In-Reply-To: <20090825224848.666530ef@whitehouse.id.au>
References: <20090825224848.666530ef@whitehouse.id.au>
Message-ID: <befaa7aa0908250913g2ca8ceceme872d46e67fcd31c@mail.gmail.com>

Wow, you even documented your change. Much thanks. I've pushed the
patch to the repository.



From sebyte at smolny.plus.com  Thu Aug 27 19:19:30 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Thu, 27 Aug 2009 19:19:30 +0000
Subject: [postmodern-devel] Authentication
Message-ID: <prah83p9.fsf@vps203.linuxvps.org>

Hi all,

It seems it is not possible to connect to a local postgres server using the
'ident sameuser' authentication method - the two functions 'connect' and
'connect-toplevel' both require user and password arguments.

I thought perhaps passing empty strings as user and password arguments might be
the way to say "use the ident sameuser method" but alas, it raises the error:

 "Database error 28000: no PostgreSQL user name specified in startup packet"

Perhaps this is a feature that is lacking, or am I missing something?

Regards,

Sebastian

P.S.  Apologies if this is a FAQ.  A Gmane search of this list for 'ident
      sameuser' returned no results.

-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From marijnh at gmail.com  Thu Aug 27 21:24:43 2009
From: marijnh at gmail.com (Marijn Haverbeke)
Date: Thu, 27 Aug 2009 23:24:43 +0200
Subject: [postmodern-devel] Authentication
In-Reply-To: <prah83p9.fsf@vps203.linuxvps.org>
References: <prah83p9.fsf@vps203.linuxvps.org>
Message-ID: <befaa7aa0908271424u62febb42te8bc807f4d880003@mail.gmail.com>

Hi Sebastian,

This is not a FAQ. Postmodern currently only implements MD5-password
authentication. I never really looked into the other methods that
Postgres supports. If you want to implement other methods, see
cl-postgres/messages.lisp and
http://www.postgresql.org/docs/current/interactive/protocol-flow.html#AEN79753
. It doesn't say anything about 'sameuser' there though, so I'm not
sure what the mechanism for that is.

Cheers,
Marijn



From attila.lendvai at gmail.com  Thu Aug 27 22:35:19 2009
From: attila.lendvai at gmail.com (Attila Lendvai)
Date: Fri, 28 Aug 2009 00:35:19 +0200
Subject: [postmodern-devel] Authentication
In-Reply-To: <befaa7aa0908271424u62febb42te8bc807f4d880003@mail.gmail.com>
References: <prah83p9.fsf@vps203.linuxvps.org>
	<befaa7aa0908271424u62febb42te8bc807f4d880003@mail.gmail.com>
Message-ID: <b54b34af0908271535j6fd4216agb9af20dd6d72f82a@mail.gmail.com>

> . It doesn't say anything about 'sameuser' there though, so I'm not
> sure what the mechanism for that is.


i think you cannot use that over the tcp connection cl-postgres
supports, only through some C interface which probably goes through
some unix sockets on the same machine.

-- 
 attila



From sabra.crolleton at gmail.com  Fri Aug 28 05:21:39 2009
From: sabra.crolleton at gmail.com (Sabra Crolleton)
Date: Thu, 27 Aug 2009 22:21:39 -0700
Subject: [postmodern-devel] Using get-dao to somehow pull in foreign key
	linked fields?
Message-ID: <200908272221.39752.sabra.crolleton@gmail.com>

I was just going to display the fields in a record and my first thought was to use the relevant class and then just call get-dao and iterate over the slots, something like:

         (item (postmodern:get-dao 'countrynotes (parse-integer id))))

However, looking at the countrynotes definition, you quickly notice that it has a foreign key:

; shortened version of the countrynotes class
(defclass countrynotes ()
  (
   (id :accessor id :col-type int4 :initarg :id )
   (country-id :accessor country-id :col-type int4 :initarg :country-id :foreign-key (countries id))
   (name :accessor name :col-type varchar :initarg :name )
   (updated-at :accessor updated-at :col-type timestamp :initarg :updated-at ))
  (:metaclass postmodern:dao-class)
    (:table-name countrynotes)(:keys id))

In displaying the countrynote, I don't really want to show the country-id, I really want to display the name of the country instead, which is another table:
(defclass countries ()
  (
   (id :accessor id :col-type int4 :initarg :id )
   (name :accessor name :col-type varchar :initarg :name )
   (updated-at :accessor updated-at :col-type timestamp :initarg :updated-at ))
  (:metaclass postmodern:dao-class)
    (:table-name countries)(:keys id))

I can easily write out the full required sql statement, but is there a simple way of just using the dao?

Sabra



From marijnh at gmail.com  Fri Aug 28 06:34:19 2009
From: marijnh at gmail.com (Marijn Haverbeke)
Date: Fri, 28 Aug 2009 08:34:19 +0200
Subject: [postmodern-devel] Using get-dao to somehow pull in foreign key
	linked fields?
In-Reply-To: <200908272221.39752.sabra.crolleton@gmail.com>
References: <200908272221.39752.sabra.crolleton@gmail.com>
Message-ID: <befaa7aa0908272334i75610c6fxd09e18a8f0acade@mail.gmail.com>

Hi Sabra,

There's no elegant provision for implicitly joining things onto DAO
objects, but there is a hack. See
http://common-lisp.net/project/postmodern/postmodern.html#with-column-writers
. Here's the only that I ever made of it (as far as I know, it's the
only use ever made of it at all), with some names replaced to protect
all the trade secrets that are obviously present in such a thing:

(defun get-foos-by (condition)
  (with-column-writers ('actives 'has-active-bars 'shape 'shape)
    (query-dao 'foo-desc
      (:select 'foo.* (:as 'shape.name 'shape)
              (:as (:exists (:select 1 :from 'bar :where (:and (:=
'foo-id 'foo.id) 'active))) 'actives)
        :from 'foo :left-join 'shape :on (:= 'shape.id 'foo.shape-id)
        :where (:raw condition)))))

Hope that helps,
Marijn



From sebyte at smolny.plus.com  Fri Aug 28 23:16:41 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Fri, 28 Aug 2009 23:16:41 +0000
Subject: [postmodern-devel] Wildacrds in string patterns
Message-ID: <k50na5ra.fsf@vps203.linuxvps.org>

Hi all,

What's the trick for using wildcards such as '%' in string patterns in queries?

I can't find anything on it in the S-SQL reference and I've tried several
different things - I either get a database error or the Emacs debugger kicks in
complaining about an end of file, or some such problem.

Regards,

Sebastian
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From info at jensteich.de  Sat Aug 29 06:58:19 2009
From: info at jensteich.de (Jens Teich)
Date: Sat, 29 Aug 2009 08:58:19 +0200
Subject: [postmodern-devel] Wildacrds in string patterns
In-Reply-To: <k50na5ra.fsf@vps203.linuxvps.org>
References: <k50na5ra.fsf@vps203.linuxvps.org>
Message-ID: <4A98D18B.3030200@jensteich.de>

Sebastian Tennant schrieb:
> Hi all,
> 
> What's the trick for using wildcards such as '%' in string patterns in queries?
> 
> I can't find anything on it in the S-SQL reference and I've tried several
> different things - I either get a database error or the Emacs debugger kicks in
> complaining about an end of file, or some such problem.


(:where (:like 'field "String%"))

Jens



From sebyte at smolny.plus.com  Sat Aug 29 08:47:32 2009
From: sebyte at smolny.plus.com (Sebastian Tennant)
Date: Sat, 29 Aug 2009 08:47:32 +0000
Subject: [postmodern-devel] Wildacrds in string patterns
References: <k50na5ra.fsf@vps203.linuxvps.org> <4A98D18B.3030200@jensteich.de>
Message-ID: <ab1j0zx7.fsf@vps203.linuxvps.org>

Quoth Jens Teich <info at jensteich.de>:
> Sebastian Tennant schrieb:
>> What's the trick for using wildcards such as '%' in string patterns in
>> queries?

> (:where (:like 'field "String%"))

Thanks Jens.  I tried the above.  It turns out the problem I am having has to
do with SLIME and non-ASCII characters in the results.

Regards,

Sebastian
-- 
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap




From khaelin at gmail.com  Sun Aug 30 13:28:25 2009
From: khaelin at gmail.com (Nicolas Martyanoff)
Date: Sun, 30 Aug 2009 15:28:25 +0200
Subject: [postmodern-devel] Large objects support
Message-ID: <20090830132824.GB1501@valhala.gaia.loc>

Hi,

Is there a large object? support in development, or do you intend to add one in
the future?

1. http://www.postgresql.org/docs/8.4/static/largeobjects.html

Regards,

-- 
Nicolas Martyanoff
   http://codemore.org
   khaelin at gmail.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/postmodern-devel/attachments/20090830/9c3cf43c/attachment.sig>

From marijnh at gmail.com  Sun Aug 30 13:49:18 2009
From: marijnh at gmail.com (Marijn Haverbeke)
Date: Sun, 30 Aug 2009 15:49:18 +0200
Subject: [postmodern-devel] Large objects support
In-Reply-To: <20090830132824.GB1501@valhala.gaia.loc>
References: <20090830132824.GB1501@valhala.gaia.loc>
Message-ID: <befaa7aa0908300649q55e93917m566519da032a18cd@mail.gmail.com>

Hi Nicolas,

Postmodern does not currently support this. There isn't a single
mention of large objects in documentation of the Postgres
Frontend/Backend protocol, so I'm not even sure how they are
implemented by the C client lib. They seem to be a hack on top of
tables containing binary arrays
(http://www.postgresql.org/docs/8.4/static/lo-implementation.html).

If you have a need for this, you can have a shot at implementing it.
Myself, I'm in 'maintenance' mode when it comes to Postgres -- I'm
happy to fix bugs and answer questions, but I don't have time to add
substantial new features.

Best,
Marijn



From khaelin at gmail.com  Sun Aug 30 15:22:22 2009
From: khaelin at gmail.com (Nicolas Martyanoff)
Date: Sun, 30 Aug 2009 17:22:22 +0200
Subject: [postmodern-devel] Large objects support
In-Reply-To: <befaa7aa0908300649q55e93917m566519da032a18cd@mail.gmail.com>
References: <20090830132824.GB1501@valhala.gaia.loc>
	<befaa7aa0908300649q55e93917m566519da032a18cd@mail.gmail.com>
Message-ID: <20090830152222.GC1501@valhala.gaia.loc>

On 09-08-30 15:49, Marijn Haverbeke wrote:
> Hi Nicolas,
> 
> Postmodern does not currently support this. There isn't a single
> mention of large objects in documentation of the Postgres
> Frontend/Backend protocol, so I'm not even sure how they are
> implemented by the C client lib. They seem to be a hack on top of
> tables containing binary arrays
> (http://www.postgresql.org/docs/8.4/static/lo-implementation.html).
> 
> If you have a need for this, you can have a shot at implementing it.
> Myself, I'm in 'maintenance' mode when it comes to Postgres -- I'm
> happy to fix bugs and answer questions, but I don't have time to add
> substantial new features.

No problem :)

If I need them, I'll try to add the support to postmodern (excellent library
by the way).

Regards,

-- 
Nicolas Martyanoff
   http://codemore.org
   khaelin at gmail.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/postmodern-devel/attachments/20090830/cffee3e7/attachment.sig>

From khaelin at gmail.com  Sun Aug 30 17:26:36 2009
From: khaelin at gmail.com (khaelin at gmail.com)
Date: Sun, 30 Aug 2009 10:26:36 -0700 (PDT)
Subject: [postmodern-devel] darcs patch: add support for the currval() and
	lastval() functions
Message-ID: <4a9ab64c.0508d00a.392c.ffffa3d6@mx.google.com>

Sun Aug 30 19:25:12 CEST 2009  khaelin at gmail.com
  * add support for the currval() and lastval() functions
-------------- next part --------------
A non-text attachment was scrubbed...
Name: add-support-for-the-currval__-and-lastval__-functions.dpatch
Type: text/x-darcs-patch
Size: 29436 bytes
Desc: A darcs patch for your repository!
URL: <https://mailman.common-lisp.net/pipermail/postmodern-devel/attachments/20090830/113d3eea/attachment.bin>

From marijnh at gmail.com  Sun Aug 30 17:39:19 2009
From: marijnh at gmail.com (Marijn Haverbeke)
Date: Sun, 30 Aug 2009 19:39:19 +0200
Subject: [postmodern-devel] darcs patch: add support for the currval()
	and lastval() functions
In-Reply-To: <4a9ab64c.0508d00a.392c.ffffa3d6@mx.google.com>
References: <4a9ab64c.0508d00a.392c.ffffa3d6@mx.google.com>
Message-ID: <befaa7aa0908301039m186b07aew24f0ede0ae750de3@mail.gmail.com>

Hi,

You'll notice that, even without your patch, something like (:currval
"foo") expands to currval('foo') -- it's just a function, and when no
special syntax is defined for a keyword, S-SQL will expand it as a
function. Given this, does the patch still serve a purpose?

Best,
Marijn



From khaelin at gmail.com  Sun Aug 30 17:55:28 2009
From: khaelin at gmail.com (Nicolas Martyanoff)
Date: Sun, 30 Aug 2009 19:55:28 +0200
Subject: [postmodern-devel] darcs patch: add support for the currval()
 and lastval() functions
In-Reply-To: <befaa7aa0908301039m186b07aew24f0ede0ae750de3@mail.gmail.com>
References: <4a9ab64c.0508d00a.392c.ffffa3d6@mx.google.com>
	<befaa7aa0908301039m186b07aew24f0ede0ae750de3@mail.gmail.com>
Message-ID: <20090830175528.GD1501@valhala.gaia.loc>

On 09-08-30 19:39, Marijn Haverbeke wrote:
> Hi,
> 
> You'll notice that, even without your patch, something like (:currval
> "foo") expands to currval('foo') -- it's just a function, and when no
> special syntax is defined for a keyword, S-SQL will expand it as a
> function. Given this, does the patch still serve a purpose?

I don't know if I do it correctly, but without my patch:
    * (s-sql:sql-compile '(:select (:currval players_id_seq)))

    "(SELECT currval(players_id_seq))"

With my patch:
    * (s-sql:sql-compile '(:select (:currval players_id_seq)))

    "(SELECT currval(E'players_id_seq'))"

It seems that the default function expansion doesn't escape object names.

Since I'm here, note that there's also a setval() function, a bit more complex
since it takes arguments (two or three), but I didn't figure how to add it
(sorry, lisp beginner here). I'll try to add it... one day.

Regards,

-- 
Nicolas Martyanoff
   http://codemore.org
   khaelin at gmail.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/postmodern-devel/attachments/20090830/78b5d25a/attachment.sig>

From marijnh at gmail.com  Sun Aug 30 18:03:34 2009
From: marijnh at gmail.com (Marijn Haverbeke)
Date: Sun, 30 Aug 2009 20:03:34 +0200
Subject: [postmodern-devel] darcs patch: add support for the currval()
	and lastval() functions
In-Reply-To: <20090830175528.GD1501@valhala.gaia.loc>
References: <4a9ab64c.0508d00a.392c.ffffa3d6@mx.google.com>
	<befaa7aa0908301039m186b07aew24f0ede0ae750de3@mail.gmail.com>
	<20090830175528.GD1501@valhala.gaia.loc>
Message-ID: <befaa7aa0908301103l77a6ee67l3ff3c8752f150192@mail.gmail.com>

> ? ?* (s-sql:sql-compile '(:select (:currval players_id_seq)))
> ? ?"(SELECT currval(players_id_seq))"

> ? ?* (s-sql:sql-compile '(:select (:currval players_id_seq)))
> ? ?"(SELECT currval(E'players_id_seq'))"

> It seems that the default function expansion doesn't escape object names.

The Postgresql currval/nextval functions take strings. Passing a
symbol, in S-SQL means passing an identifier. If you are generating
s-sql expressions, have a symbol where you need a string, you can just
call to-sql-name on it. Doing it that way should also make calling
setval easier.

Best,
Marijn



From khaelin at gmail.com  Mon Aug 31 08:10:52 2009
From: khaelin at gmail.com (Nicolas Martyanoff)
Date: Mon, 31 Aug 2009 10:10:52 +0200
Subject: [postmodern-devel] darcs patch: add support for the currval()
 and lastval() functions
In-Reply-To: <befaa7aa0908301103l77a6ee67l3ff3c8752f150192@mail.gmail.com>
References: <4a9ab64c.0508d00a.392c.ffffa3d6@mx.google.com>
	<befaa7aa0908301039m186b07aew24f0ede0ae750de3@mail.gmail.com>
	<20090830175528.GD1501@valhala.gaia.loc>
	<befaa7aa0908301103l77a6ee67l3ff3c8752f150192@mail.gmail.com>
Message-ID: <20090831081052.GA1724@valhala.gaia.loc>

On 09-08-30 20:03, Marijn Haverbeke wrote:
> > ? ?* (s-sql:sql-compile '(:select (:currval players_id_seq)))
> > ? ?"(SELECT currval(players_id_seq))"
> 
> > ? ?* (s-sql:sql-compile '(:select (:currval players_id_seq)))
> > ? ?"(SELECT currval(E'players_id_seq'))"
> 
> > It seems that the default function expansion doesn't escape object names.
> 
> The Postgresql currval/nextval functions take strings. Passing a
> symbol, in S-SQL means passing an identifier. If you are generating
> s-sql expressions, have a symbol where you need a string, you can just
> call to-sql-name on it. Doing it that way should also make calling
> setval easier.

You seem to be right :) Perhaps should nextval() be removed then, for
consistency purpose.

Regards,

-- 
Nicolas Martyanoff
   http://codemore.org
   khaelin at gmail.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/postmodern-devel/attachments/20090831/43ad9f02/attachment.sig>

From marijnh at gmail.com  Mon Aug 31 08:17:58 2009
From: marijnh at gmail.com (Marijn Haverbeke)
Date: Mon, 31 Aug 2009 10:17:58 +0200
Subject: [postmodern-devel] darcs patch: add support for the currval()
	and lastval() functions
In-Reply-To: <20090831081052.GA1724@valhala.gaia.loc>
References: <4a9ab64c.0508d00a.392c.ffffa3d6@mx.google.com>
	<befaa7aa0908301039m186b07aew24f0ede0ae750de3@mail.gmail.com>
	<20090830175528.GD1501@valhala.gaia.loc>
	<befaa7aa0908301103l77a6ee67l3ff3c8752f150192@mail.gmail.com>
	<20090831081052.GA1724@valhala.gaia.loc>
Message-ID: <befaa7aa0908310117x6af19724w91f88d213eb18dc9@mail.gmail.com>

Ah, there's already a :nextval. Now I see your point. I guess someone
(possibly me) thought that was a good idea at the time. My issue is,
though, that there's so much built-in functions in Postgres that
providing a convenience wrapper for all of them would make the library
huge.

I'd remove nextval, but that might break someone's code. The
consistency isn't too bad, since the definition of :nextval also
allows strings as input, people passing these functions strings (the
logical thing to do) will never even notice this.

Best,
Marijn