From scaekenberghe at common-lisp.net  Tue Apr 24 14:42:41 2007
From: scaekenberghe at common-lisp.net (Sven Van Caekenberghe)
Date: Tue, 24 Apr 2007 16:42:41 +0200
Subject: [cl-sockets-devel] Re: Status of cl-soap project
In-Reply-To: <20070423105816.GD22899@merkur.math.uni-magdeburg.de>
References: <20070412094026.GH22899@merkur.math.uni-magdeburg.de>
	<FB9CA2BA-3E16-4A37-B13C-00651DDF9D54@mac.com>
	<20070417141624.GS22899@merkur.math.uni-magdeburg.de>
	<E0606E55-7973-41E1-AD45-102937B3DE23@mac.com>
	<20070423105816.GD22899@merkur.math.uni-magdeburg.de>
Message-ID: <97414A9D-CAAC-49DF-844E-4E2743AD26D6@common-lisp.net>

Utz,

On 23 Apr 2007, at 12:58, Utz-Uwe Haus wrote:

> On the positive side: I can now successfully parse (including schema
> <import> and <include> the wsdl file and make a wsdl-based soap call.
> This involves handling multiple schemata in the wsdl type slots and
> handling anonymous sequences inside complexType elements, as well  
> as the
> xsd:any tag. The latter two only on the parsing/template side, since:
> On the negative side I have to admit that the template-based xsd
> handling is reaching its limits on the above xml -- I run into a stack
> overflow because of recursive type resolution when generating the
> (infinite) template.
> It seems one should get by with generating only a single layer of
> template code whenever necessary, a little ugly but better than a  
> total
> rewrite.

That is great news! You are really taking CL-SOAP forward.

> If you want some patches for the current state let me know. Do you  
> have
> an idea about a set of automatic tests (like for the google api -- but
> with which token?) to perform to find out whether I broke anything  
> that
> worked before?

Yes, we need to make sure that we (you) do not break the older  
functionality.
But that is difficult to do for the reasons you mentioned.

> When the template problem is fixed I'll try to come up with some
> autogeneration of CLOS classes for the wsdl contents -- that's what I
> really want to use in my project.

Even better, if I recall correctly that is more or less where we left  
off.

I have taken the liberty of CC'ing this message to the CL-SOAP-DEVEL  
list,
because I think it is important that keep a record and potentially  
involve others.

I hope Alain Picard reads this too.

Regards,

Sven



From haus+cl-soap at mail.math.uni-magdeburg.de  Tue Apr 24 19:02:00 2007
From: haus+cl-soap at mail.math.uni-magdeburg.de (Utz-Uwe Haus)
Date: Tue, 24 Apr 2007 21:02:00 +0200
Subject: [cl-sockets-devel] Re: [cl-soap-devel] Re: Status of cl-soap project
In-Reply-To: <97414A9D-CAAC-49DF-844E-4E2743AD26D6@common-lisp.net>
References: <20070412094026.GH22899@merkur.math.uni-magdeburg.de>
	<FB9CA2BA-3E16-4A37-B13C-00651DDF9D54@mac.com>
	<20070417141624.GS22899@merkur.math.uni-magdeburg.de>
	<E0606E55-7973-41E1-AD45-102937B3DE23@mac.com>
	<20070423105816.GD22899@merkur.math.uni-magdeburg.de>
	<97414A9D-CAAC-49DF-844E-4E2743AD26D6@common-lisp.net>
Message-ID: <20070424190200.GN22899@merkur.math.uni-magdeburg.de>

On Tue, Apr 24, 2007 at 04:42:41PM +0200, Sven Van Caekenberghe wrote:
> Yes, we need to make sure that we (you) do not break the older  
> functionality.
> But that is difficult to do for the reasons you mentioned.

I'll see if there is some developer-token program or the like. I seem to
have something like that for amazon's program around somewhere.
 
> >When the template problem is fixed I'll try to come up with some
> >autogeneration of CLOS classes for the wsdl contents -- that's what I
> >really want to use in my project.
> 
> Even better, if I recall correctly that is more or less where we left  
> off.

Ok. I have fixed the template stuff by wrapping the deeper layers of the
template into continuations that can be evaluated as needed. That
actually is speeding things up for my complex xml types.

The last problem I see for my application before attacking CLOS-wrapping
is now that the namespace support is broken if the result is typed with
namespace-qualified elements outside the |s0|-namespace used for the
soap reply. Should not be too hard to fix.

> I have taken the liberty of CC'ing this message to the CL-SOAP-DEVEL  
> list,
> because I think it is important that keep a record and potentially  
> involve others.

... therefore I continued on the list ...

Later,
Utz

-- 
Utz-Uwe Haus                               haus at mail.math.uni-magdeburg.de
Inst. f. Mathemat. Optim.                                    utz at uuhaus.de
Uni Magdeburg                    PGP keys  1024/6AD23BE1 and 2048/5D0B72A1
GERMANY                          available via keyservers or email request 


From haus+cl-soap at mail.math.uni-magdeburg.de  Tue Apr 24 19:10:28 2007
From: haus+cl-soap at mail.math.uni-magdeburg.de (Utz-Uwe Haus)
Date: Tue, 24 Apr 2007 21:10:28 +0200
Subject: [cl-sockets-devel] Re: [cl-soap-devel] Re: Status of cl-soap
	project
In-Reply-To: <20070424190200.GN22899@merkur.math.uni-magdeburg.de>
References: <20070412094026.GH22899@merkur.math.uni-magdeburg.de>
	<FB9CA2BA-3E16-4A37-B13C-00651DDF9D54@mac.com>
	<20070417141624.GS22899@merkur.math.uni-magdeburg.de>
	<E0606E55-7973-41E1-AD45-102937B3DE23@mac.com>
	<20070423105816.GD22899@merkur.math.uni-magdeburg.de>
	<97414A9D-CAAC-49DF-844E-4E2743AD26D6@common-lisp.net>
	<20070424190200.GN22899@merkur.math.uni-magdeburg.de>
Message-ID: <20070424191028.GO22899@merkur.math.uni-magdeburg.de>

(following up on my own posting)

On Tue, Apr 24, 2007 at 09:02:00PM +0200, Utz-Uwe Haus wrote:
> On Tue, Apr 24, 2007 at 04:42:41PM +0200, Sven Van Caekenberghe wrote:
> > Yes, we need to make sure that we (you) do not break the older  
> > functionality.
> > But that is difficult to do for the reasons you mentioned.
> 
> I'll see if there is some developer-token program or the like.

There seems to be a sandbox environment available that could be a start:
<http://www.google.com/apis/adwords/developer/adwords_api_sandbox.html>

If someone has a typical soap-calls for the AdWords API at hand we could
set up a testsuite from that which does not require a true developer
token and does not incur costs. Unfortunately I have no idea what
exactly to test in that API...

> > I have taken the liberty of CC'ing this message to the CL-SOAP-DEVEL  
> > list, because I think it is important that keep a record and
> > potentially  involve others.

Why are we getting [cl-sockets-devel] subject tags thrown in from the
mailman?

later,
Utz

-- 
Utz-Uwe Haus                               haus at mail.math.uni-magdeburg.de
Inst. f. Mathemat. Optim.                                    utz at uuhaus.de
Uni Magdeburg                    PGP keys  1024/6AD23BE1 and 2048/5D0B72A1
GERMANY                          available via keyservers or email request 


From haus+cl-soap at mail.math.uni-magdeburg.de  Fri Apr 27 21:41:00 2007
From: haus+cl-soap at mail.math.uni-magdeburg.de (Utz-Uwe Haus)
Date: Fri, 27 Apr 2007 23:41:00 +0200
Subject: [cl-sockets-devel] Progress report on cl-soap extensions
Message-ID: <20070427214059.GY22899@merkur.math.uni-magdeburg.de>

Dear listmembers,

just letting you know that I successfully managed to query the
NCBI efetch soap service as described on
    <http://www.ncbi.nlm.nih.gov/entrez/query/static/esoap_help.html>
It seems to be one of the more complicated service definitions around
-- around 5000 schema elements (imported and included), recursive types,
ref elements that need late binding and have overridden attributes
etc. Unfortunately the wsdl files are buggy -- I am in contact with the
maintainers at the NIH.  The errors are actually found by cl-soap :)
(database results cannot be parsed with the given specification).

There remains a lot to be done.  Mainly, I would like any opinion you
may have on two major design issues that I need to address:

1) We will need proper namespace support -- the above wsdl uses 8
namespaces and schemata. This works quite nicely, but in the
parse-results we currently return strings as element names, and they do
not have namespace prefixes. Obviously this makes the output ambiguous
if the same name appears in more than one namespace.

I would like to fix this by returning (namespace-package-qualified-) symbols;
then we could map to print-name for existing code that expects strings.

2) In line with (1) and my goal to automatically generate lisp code for
the soap interface I would like to stuff always collect xml names in the
proper namespace's package and keep a global dictionary of schema
elements (and namespaces) in the cl-soap package. This sounds worse than
it is -- actually Allegro SOAP does just this (see soap-define-element
etc in their package).

If nobody objects I'll go ahead. If anybody wants to see the current
code let me know -- no public repository at hand, currently.

later,
Utz
-- 
Utz-Uwe Haus                               haus at mail.math.uni-magdeburg.de
Inst. f. Mathemat. Optim.                                    utz at uuhaus.de
Uni Magdeburg                    PGP keys  1024/6AD23BE1 and 2048/5D0B72A1
GERMANY                          available via keyservers or email request 


From alexander.kjeldaas at gmail.com  Mon Apr 30 13:07:00 2007
From: alexander.kjeldaas at gmail.com (Alexander Kjeldaas)
Date: Mon, 30 Apr 2007 15:07:00 +0200
Subject: [cl-sockets-devel] Progress report on cl-soap extensions
In-Reply-To: <20070427214059.GY22899@merkur.math.uni-magdeburg.de>
References: <20070427214059.GY22899@merkur.math.uni-magdeburg.de>
Message-ID: <b2ec2dee0704300607g4e8e5c74t758fce19c87cc63c@mail.gmail.com>

On 4/27/07, Utz-Uwe Haus <haus+cl-soap at mail.math.uni-magdeburg.de> wrote:
>
> If nobody objects I'll go ahead. If anybody wants to see the current
> code let me know -- no public repository at hand, currently.
>

Just to point you to a patch to cl-soap I did a while ago.  It made it
possible for me to query some fairly complex government soap-services.
  Take a look if you are interested:

http://common-lisp.net/pipermail/cl-soap-devel/2006-August/000031.html

astor
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cl-soap.patch.gz
Type: application/x-gzip
Size: 7769 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/cl-soap-devel/attachments/20070430/0c983488/attachment.bin>

From haus+cl-soap at mail.math.uni-magdeburg.de  Mon Apr 30 14:05:02 2007
From: haus+cl-soap at mail.math.uni-magdeburg.de (Utz-Uwe Haus)
Date: Mon, 30 Apr 2007 16:05:02 +0200
Subject: [cl-sockets-devel] Progress report on cl-soap extensions
In-Reply-To: <b2ec2dee0704300607g4e8e5c74t758fce19c87cc63c@mail.gmail.com>
References: <20070427214059.GY22899@merkur.math.uni-magdeburg.de>
	<b2ec2dee0704300607g4e8e5c74t758fce19c87cc63c@mail.gmail.com>
Message-ID: <20070430140502.GC22899@merkur.math.uni-magdeburg.de>

Hi,

On Mon, Apr 30, 2007 at 03:07:00PM +0200, Alexander Kjeldaas wrote:
> Just to point you to a patch to cl-soap I did a while ago.  It made it
> possible for me to query some fairly complex government soap-services.

I've actually seen that -- after starting -- and have it on my todo
list. Some things like defaulting are still missing from my version, but
others I had to solve differently.

later,
Utz


-- 
Utz-Uwe Haus                               haus at mail.math.uni-magdeburg.de
Inst. f. Mathemat. Optim.                                    utz at uuhaus.de
Uni Magdeburg                    PGP keys  1024/6AD23BE1 and 2048/5D0B72A1
GERMANY                          available via keyservers or email request