From thomas at src.no  Tue Feb 15 12:01:32 2005
From: thomas at src.no (Thomas Stenhaug)
Date: Tue, 15 Feb 2005 13:01:32 +0100
Subject: [Cl-store-devel] Patch for more comprehensive handling of packages
Message-ID: <4211E49C.3010109@src.no>

I make and manipulate packages at runtime, so I needed to serialize and 
deserialize their state.  That's what the patch is for.

cl-store is great stuff, btw. :)


Thomas
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: more-comprehensive-package-handling.diff
URL: <https://mailman.common-lisp.net/pipermail/cl-store-devel/attachments/20050215/b1120c05/attachment.ksh>

From sdr at jhb.ucs.co.za  Wed Feb 16 13:27:21 2005
From: sdr at jhb.ucs.co.za (Sean Ross)
Date: 16 Feb 2005 15:27:21 +0200
Subject: [Cl-store-devel] Patch for more comprehensive handling of packages
References: <4211E49C.3010109@src.no>
Message-ID: <87u0oc63eu.fsf@sdr.ucs.co.za>

Thomas Stenhaug <thomas at src.no> writes:

> I make and manipulate packages at runtime, so I needed to serialize
> and deserialize their state.  That's what the patch is for.
> 
> cl-store is great stuff, btw. :)
> 
> 
> Thomas

Hi,

 Thanks for the patch.
 
 I've applied it to cvs with a few modifications. The gist of the 
 patch is the same but I've changed external-symbols and internal-symbols
 to return a vector since lispworks can choke when restoring large lists,
 internal-symbols is also a little more choosy and will ignore internal
 symbols which came from packages on it's use list. I also added 
 package-shadowing-symbols to the stored values.

 The last change was a variable which determines whether to store
 the packages on the use list as full packages or as a package-name.
 
 BTW, I was reading through old cmucl archives and came across a post
 from Erik Enge which said that you were writing cl-pile which has 
 'transactions and several other neat features' so I was wondering 
 if you have any ideas for new features for cl-store. 

- Sean

-- 
"My doctor says that I have a malformed public-duty gland and a
 natural  deficiency in moral fibre," he muttered to himself, "and
 that I am therefore excused from saving Universes."
 - Life, the Universe, and Everything     Douglas Adams.



From thomas at src.no  Wed Feb 16 13:55:45 2005
From: thomas at src.no (Thomas Stenhaug)
Date: Wed, 16 Feb 2005 14:55:45 +0100
Subject: [Cl-store-devel] Patch for more comprehensive handling of packages
In-Reply-To: <87u0oc63eu.fsf@sdr.ucs.co.za>
References: <4211E49C.3010109@src.no> <87u0oc63eu.fsf@sdr.ucs.co.za>
Message-ID: <421350E1.6060803@src.no>

Sean Ross wrote:

> 
> I've applied it to cvs with a few modifications. 
>
Great stuff!  Thanks.

> 
> BTW, I was reading through old cmucl archives and came across a post
> from Erik Enge which said that you were writing cl-pile which has 
> 'transactions and several other neat features' so I was wondering 
> if you have any ideas for new features for cl-store. 
>  
>
What I'm trying to do is actually to replace my own serialization-code 
in cl-pile with cl-store, since cl-store is more comprehensive, faster, 
better tested and documented. :)

cl-pile is an implementation of the prevalance [1] idea, with some MOP 
extensions for giving classes, objects and slots things like title, 
description, creation-date etc.

If the experiment of using cl-store for the serialization-layer, I'll 
try to seperate out the prevalence-parts of it, and let you look at it, 
to see if it's of interest.

----
[1] http://www-106.ibm.com/developerworks/library/wa-objprev/




From thomas at src.no  Wed Feb 16 15:22:34 2005
From: thomas at src.no (Thomas Stenhaug)
Date: Wed, 16 Feb 2005 16:22:34 +0100
Subject: [Cl-store-devel] Making GET-SLOT-DETAILS a generic
Message-ID: <4213653A.1080100@src.no>

To be able to serialize non-standard slots, I made the GET-SLOT-DETAILS 
into a generic funtion, and exported it, so I can specialize on it.  
Worked like a charm for me.


Thomas
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: generic-get-slot-details.diff
URL: <https://mailman.common-lisp.net/pipermail/cl-store-devel/attachments/20050216/0f6c1a48/attachment.ksh>

From sdr at jhb.ucs.co.za  Thu Feb 17 10:04:09 2005
From: sdr at jhb.ucs.co.za (Sean Ross)
Date: 17 Feb 2005 12:04:09 +0200
Subject: [Cl-store-devel] Making GET-SLOT-DETAILS a generic
References: <4213653A.1080100@src.no>
Message-ID: <87hdkbbizq.fsf@sdr.ucs.co.za>

Thomas Stenhaug <thomas at src.no> writes:

> To be able to serialize non-standard slots, I made the
> GET-SLOT-DETAILS into a generic funtion, and exported it, so I can
> specialize on it.  Worked like a charm for me.

Thanks, 

 I've applied your patch (with a few conditialized statements to 
 handle ecl and clisp) to cvs. 

 I've also added the GF serializable-slots since most of the time
 you want to do custom serialization of objects is when you want
 to remove various slots from the list of slots to serialize 
 (at least thats what I found)

 so no you can do 
 
 (defclass foo () 
   ((a :accessor a :initarg :a)
    (b :accessor b :initarg :b)))
 
 (defmethod serializable-slots ((object foo))
   (remove 'b (call-next-method) :key 'slot-definition-name))
 
 to only serialize slot A.


 I guess you could do this with a metaclass and specializing
 compute-slots but that seems like a bit of a pain.
 
 

 Thanks for the patch.

-- Sean

-- 
"My doctor says that I have a malformed public-duty gland and a
 natural  deficiency in moral fibre," he muttered to himself, "and
 that I am therefore excused from saving Universes."
 - Life, the Universe, and Everything     Douglas Adams.



From thomas at src.no  Thu Feb 17 10:51:57 2005
From: thomas at src.no (Thomas Stenhaug)
Date: Thu, 17 Feb 2005 11:51:57 +0100
Subject: [Cl-store-devel] Making GET-SLOT-DETAILS a generic
In-Reply-To: <87hdkbbizq.fsf@sdr.ucs.co.za>
References: <4213653A.1080100@src.no> <87hdkbbizq.fsf@sdr.ucs.co.za>
Message-ID: <4214774D.5060209@src.no>

Sean Ross wrote:

> (defclass foo () 
>   ((a :accessor a :initarg :a)
>    (b :accessor b :initarg :b)))
> 
> (defmethod serializable-slots ((object foo))
>   (remove 'b (call-next-method) :key 'slot-definition-name))
> 
> to only serialize slot A.
>  
>
Great!  FWIW, that's the same solution I used in my 
serialization-library too.


Thomas




From sdr at jhb.ucs.co.za  Fri Feb 18 12:44:23 2005
From: sdr at jhb.ucs.co.za (Sean Ross)
Date: 18 Feb 2005 14:44:23 +0200
Subject: [Cl-store-devel] [Announce] CL-STORE 0.5
Message-ID: <87braiujfc.fsf@sdr.ucs.co.za>


CL-STORE 0.5 has just been uploaded to cl.net and
can now be downloaded or asdf-installed.


Noteworthy Changes
======================
* New Magic Number, breaking backward compatibility.
* Support for storing structure definitions with CMUCL.
* cl-store-xml.asd has been removed.
* Bug fix on hash-table restoration.
* Argument order on most GF's have been reorganized to 
  have the backend first.
* fix-clisp.lisp renamed to mop.lisp
* Support added for ECL.
* resolving-object and setting have been reworked allowing 
  setting be used like setf, *postfix-setters* has been 
  removed.
* Backend designators have been added allowing the backend
  argument to store, restore and with-backend to be either
  a backend-object or a backend name.
* defrestore-? macros now expand to generic-functions so
  these macros now accept optional qualifier arguments.
* stream-type of backends have been changed from char and 
  binary to anything acceptable as an element-type argument
  to open.
* multiple-value-store has been removed.
* Floating point numbers are now portable between implementations 
  while still supporting Nan floats and friends.
* Bug Fix for standard-class serialization from Thomas Stenhaug.
* Packages are now fully stored (Thanks to Thomas Stenhaug).
* New Exported Method, get-slot-details which can be specialized
  when store a classes slots (Thanks to Thomas Stenhaug).
* New Exported Methods, serializable-slots and 
  serializable-slots-using-class which can be used to customize
  which slots are serialized when storing clos instances.

Many apologies for any breakages.

Enjoy,
 Sean.

-- 
"My doctor says that I have a malformed public-duty gland and a
 natural  deficiency in moral fibre," he muttered to himself, "and
 that I am therefore excused from saving Universes."
 - Life, the Universe, and Everything     Douglas Adams.