modeling-users — For users: questions, support, announcements, etc.

[Modeling-users] Intro/Question

[Clay <ho...@ho...>]

Hi folks,

First off, let me say that I'm pretty impressed with what you have 
started here.  I've been in a similar situation a few times, and decided 
something like this was sorely needed, but never had the time to put 
anything quite so coherent together to address it.  So, nice work.

I'm trying to get started small, and have hit a question.

I'm trying to think of this in terms of having the 'pymodel' file be the 
central repository for schema related issues.  That seems to be the 
design intent anyway -- and really how I would like to do things 
(central location).

Suppose I'm going to implement /etc/passwd /etc/group type concepts.  
This gives 'user' 'group' and 'usergroup' entities, plus associations.  
Great.

using groupID and userID as the primary keys, and the correlation table 
is the tuple of the two.

Heres' my issue.  Clearly you want to enforce uniqueness of username 
(strings) as well as user-ids and integer primary keys.  Further, this 
is something you need the DBMS to enforce.  Since I'm effectively 
specifying my schema in pymodel, it would seem reasonable to be able to 
add this type of constraint to that specification.

Either that, or have a coherent way of maintaining them separatly in a 
non-disjoint sort of way.

Looking for either something I've missing in the modelling, or a good 
technique to handle this class of problem.

Thanks,
Clay Hopperdietzel

Re: [Modeling-users] Newbie questions

[Sebastien B. <sbi...@us...>]

    Hi Stephen,

Stephen Nesbitt <sne...@co...>:
> All:
> 
> A couple of quick (I hope) newbie questions.
> 
> First what is the canonical way to override the the database access stuff 
> embedded in the model? Essentially I've modeled our production database, but
> want to validate and test against a test database without modifying the model
> 

I'd say the canonical way is to create a file, say, test_DB.cfg,
that overwrite the necessary db-connection dict.
(host, port, user, password)

Then set the env. variable MDL_DB_CONNECTIONS_CFG
to the full path of this .cfg file, and it will be automatically used
to overwrite the specified field with the associated values
just after the model is loaded.

See http://modeling.sourceforge.net/UserGuide/env-vars-core.html
for details.

However, please do some testing before trying to drop
the whole database!! --probably the best way to avoid shooting
yourself in the foot is to remove the user/password/host/port stuff
from your model, and have two deidacted db_config.cfg file, one on
the production machine, the other one on the test machine.
This way if you forget to set the env. variable you won't hit the
production db, just get some errors.

> Second, how do I handle commits and rollbacks? I had an instance were I had a
> failed transaction and ended up with a hung transaction and no obvious way to
> roll it back.
> 
> Thanks in advance!

Could you be more specific? Do you have an example? Normally
the framework does/should do all the commit or rollback stuff
for you, it shouldn't leave a opened transaction.

What do you mean exactly by a "hung transaction"?
(BTW: which db do you use?)

If this is a bug, we'd better be able to reproduce it to get rid of it :-/

-- Sébastien.

[Modeling-users] Newbie questions

[Stephen N. <sne...@co...>]

All:

A couple of quick (I hope) newbie questions.

First what is the canonical way to override the the database access stuff 
embedded in the model? Essentially I've modeled our production database, but 
want to validate and test against a test database without modifying the model 
in advance.

Second, how do I handle commits and rollbacks? I had an instance were I had a 
failed transaction and ended up with a hung transaction and no obvious way to 
roll it back.

Thanks in advance!

-steve
-- 

Stephen Nesbitt
Senior Configuration Management Engineer
The Cobalt Group
206.219.8271
sne...@co...

Re: [Modeling-users] Update

[Sebastien B. <sbi...@us...>]

Matthew Patton <pa...@dm...> wrote:
> Hello Sebastien et al,
> 	It's been a while since Tom and I posted on this list, and we just
> wanted to let you know of our final decision to go with J2EE instead of
> trying to develop an Enterprise Application Framework in Python.  Since
> J2EE provides everything we need and there are good open source
> implementations available (JBoss, Jonas), we felt it would not be our
> place to reinvent these low-level architectural tools.  We're now
> commencing development in J2EE on a receipting system as part of a larger
> project for providing OSS software for non-profits (URL:
> http://daniel.sf.net/).  Thanks for all of your help and for the
> interesting discussions and input.


Oh well, too bad you've decided to leave the wonderful and colorful
world of python ;) Anyhow, be sure that the discussions you and Tom
triggered / participated in will eventually make the User's Guide
better one day!

  Wishing you good luck for the future and your projects, cheers,

-- S=E9bastien.

Re: [Modeling-users] Couple of questions...

[Sebastien B. <sbi...@us...>]

John Lenton <jo...@vi...> wrote:
> agreed, and I think I pointed out (I know I tried to point out) the
> second one of these. However the first is what I understood was being
> requested.

Seems like I misunderstood your post, sorry

> The way I think I'd *like* to solve the second one is along the lines of
>=20
>   ec.fetch('Writer.books.publisher', unique=3DTrue)
>=20
> or, perhaps (if something sensible can be found to fill in the
> ellipsis),
>=20
>   ec.fetch('Writer.books.publisher', groupBy=3D'...')
>=20
>=20
> The first could be ameliorated by having fetch return an iterator
> instead of a list...
[+ more on "smarter lists", such as algebra sets]

Interesting idea indeed, I'm afraid i've not enough time to comment on this
now though. Should I forget about it, you can recall me of your comments in=
 a
week when I'll be back online. Even if it sounds like dropping support for
py2.1 (not sure however), it seems bright enough to be worth the
investigation ;)

  [do not misunderstannd me here, I do not plan to drop support for
   py2.1 by now]

> References:
> 1) http://www.cs.sfu.ca/CC/354/zaiane/material/notes/Chapter3/node7.html

Still did not find the time or energy :) to look at the reference, but I
promise we'll discuss this deeper.

-- S=E9bastien.

Re: [Modeling-users] Couple of questions...

[Sebastien B. <sbi...@us...>]

Mario Ruggier <ma...@ru...> wrote:
> > Almost... because EXISTS applies to a independent subquery, while the o=
ther
> > keywords are constraints on the current query. Would this be a possible=
 source
> > of confusion?
> >=20
> > What happens if we intermix these keywords, e.g.
> >=20
> > ec.fetch('Writer', 'books.publisher.numPages>50
> > AND (books.publisher.bizName IN ["P1","P2"])
> > AND NOT EXISTS(books.publisher.bizName NOT IN ["P1","p2"] ) '
> >=20
> > to get the authors that have books with at least 51 pages as well as ha=
ving
> > published with both of P1 and P2 but with no one else. Would this work?=
 There
> > would be a simpler way to do this?

> oops, sorry, just realized that's pretty silly. Lousy example...
> Make that something like:
>=20
> ec.fetch('Writer', 'books.publisher.numPages>50
> AND (books.publisher.bizName IN ["P1","P2"])
> AND NOT EXISTS(books.publisher.country NOT IN ["c1","c2"] ) '
>=20
> or so...

Not a problem:) My idea of this is that everything should nicely play
one w/ the other, or we'll lose most of the feature, so we should be
able to intermix every keyword w/o problem. I'll play with this some
day for sure. To be more precise, in our case (a fetchspec) EXISTS do
apply to an independent query but this query is always bound to the
original one: look at this:

  =3D# SELECT DISTINCT t0.ID, t0.LAST_NAME, t0.FIRST_NAME, t0.AGE,
  -#                 t0.FK_WRITER_ID, t0.BIRTHDAY
  -# FROM WRITER t0
  -# WHERE NOT EXISTS (SELECT * FROM BOOK,WRITER,PUBLISHER
  (#                   WHERE WRITER.ID=3Dt0.ID
  (#                         AND WRITER.ID=3DBOOK.FK_WRITER_ID=20
  (#                         AND BOOK.FK_PUBLISHER=3DPUBLISHER.ID
  (#                         AND PUBLISHER.BIZ_NAME!=3D'P1');

the second WHERE clause binds the "independent query" to the first one
with WRITER.ID=3Dt0.ID.

--> that's why I think this shouldn't be difficult to implement, but I
    need to experiment with nested EXISTS clause (I do not have enonugh
    sql knowledge to foresee the possible coming complications).


  BTW I'll will be mostly off for a week, not sure if I'll read email
      for that time, and I'll be back online for sure some day next
      week.

-- S=E9bastien.

Re: [Modeling-users] Couple of questions...

[Sebastien B. <sbi...@us...>]

        Hi Aaron,

"Aaron Freeman" <afr...@em...> writes:
> First, thank you all for the fast and details replies; they are much
> appreciated. ;-)
>=20
> I've been playing around with the code that was posted, and while I still
> don't have it working, it is getting closer.
>=20

I'm sorry, I assumed py2.2+, not py2.1

We definitely still support py2.1, but of course then some of the
features are not available. For example, there's nothing like properties
in py2.1.

  The code I posted should work given that:

- you replace p.bizname=3D"x" w/ p.setBizname("o") and
              p.bizname     w/ the appropriate getter p.getBizname()

  (this will make the ValidationException at saveChanges() disappear)

- you build the package/modules/classes w/:

    >>> build(model, define_properties=3D0)

  since py2.1 has no properties.


And that's obviously what you discovered on your own.

> 1) Based on some archived posts, I was under the impression that Modeling
> core supports Python 2.1 (I am using 2.1.3 since this is leading up to a
> Zope 2.6.4 product); is that still true?

To be clear: yes, it is.

> 2) Do I need the Modeling.dynamic module?  This appears to require python
> 2.2, but if the classes are being generated via an XML file or the ZModel=
er
> tool, this shouldn't be required, right?

You absolutely do not need too, right. I just use it (even though it's
not documented yet) since most people tends to think its the easiest way
(including me :) --but that does not mean that code generation, either
from the ZModeler or from mdl_generate_python_code.py, is deprecated.


-- S=E9bastien.

[Modeling-users] Update

[Matthew P. <pa...@dm...>]

Hello Sebastien et al,
	It's been a while since Tom and I posted on this list, and we just
wanted to let you know of our final decision to go with J2EE instead of
trying to develop an Enterprise Application Framework in Python.  Since
J2EE provides everything we need and there are good open source
implementations available (JBoss, Jonas), we felt it would not be our
place to reinvent these low-level architectural tools.  We're now
commencing development in J2EE on a receipting system as part of a larger
project for providing OSS software for non-profits (URL:
http://daniel.sf.net/).  Thanks for all of your help and for the
interesting discussions and input.

						Matt

______________________________________________________________
Matthew Patton                                  pa...@dm...
DiscipleMakers Headquarters: (814)234-7975 x32

Re: [Modeling-users] Couple of questions...

[John L. <jo...@vi...>]

Sender: John Lenton <jo...@ma...>

On Tue, Apr 20, 2004 at 11:46:10PM +0200, Sebastien Bigaret wrote:
> 
>   Yes, that's another approach indeed. But while I'm a fervent defender
> for KeyValueCoding ;)) in such cases one should be careful. Here are
> some thoughts:
> 
>   - if the database have a lot of books, the KVC approach will load them
>     all in memory: KVC traverses relationships at the object level, and
>     in this case this means fetching Books published by every Publisher,
>     then these Books' Authors. This is probably the worst case one can
>     get. That's probably why you were asking for a callable vFKP on a
>     class, weren't you?
> 
>   - In the resulting lists, if an author has published N books for a
>     publisher, the corresponding list will receive N duplicates for that
>     author.

agreed, and I think I pointed out (I know I tried to point out) the
second one of these. However the first is what I understood was being
requested.

The way I think I'd *like* to solve the second one is along the lines of

  ec.fetch('Writer.books.publisher', unique=True)

or, perhaps (if something sensible can be found to fill in the
ellipsis),

  ec.fetch('Writer.books.publisher', groupBy='...')


The first could be ameliorated by having fetch return an iterator
instead of a list...

 in fact...

  if fetch returned something smarter than a list, it might be
  possible to do something like

    ec.fetch('Writer') - \
        ec.fetch('Writer', 'books.publisher.bizName == "P1"')

  this means ec.fetch returns a set, as in relational algebra[1] sets,
  which you can operate on. These set objects wouldn't perform the
  select on the database until you accessed them as sequences, so the
  above wouldn't even touch the database until someone used the result
  for something other than operating on other sets. Maybe 'fetch' is a
  bad name for a method returning something that hasn't actually
  fetched :)

  If this falls in the 'intersting, but show me the code' category,
  I'm afraid it'll have to wait :(

On Wed, Apr 21, 2004 at 10:55:26PM +0200, Mario Ruggier wrote:
> ec.fetch('Writer', 'books.publisher.numPages>50
> AND (books.publisher.bizName IN ["P1","P2"])
> AND NOT EXISTS(books.publisher.country NOT IN ["c1","c2"] ) '

well, that's the same as

    ec.fetch('Writer', 'books.publisher.numPages > 50 \
            AND books.publisher.bizName IN ["P1", "P2"]') - \
        ec.fetch('Writer', 'books.publisher.country IN ["c1","c2"]')

right?

  
References:
1) http://www.cs.sfu.ca/CC/354/zaiane/material/notes/Chapter3/node7.html

-- 
John Lenton (jo...@vi...) -- Random fortune:
Above all else -- sky.

Re: [Modeling-users] Couple of questions...

[Mario R. <ma...@ru...>]

> ec.fetch('Writer', 'books.publisher.numPages>50
> AND (books.publisher.bizName IN ["P1","P2"])
> AND NOT EXISTS(books.publisher.bizName NOT IN ["P1","p2"] ) '

oops, sorry, just realized that's pretty silly. Lousy example...
Make that something like:

ec.fetch('Writer', 'books.publisher.numPages>50
AND (books.publisher.bizName IN ["P1","P2"])
AND NOT EXISTS(books.publisher.country NOT IN ["c1","c2"] ) '

or so...

mario

RE: [Modeling-users] Couple of questions...

[Aaron F. <afr...@em...>]

First, thank you all for the fast and details replies; they are much
appreciated. ;-)

I've been playing around with the code that was posted, and while I still
don't have it working, it is getting closer.

First, I tried the Sebastien's code example (called extended.py here); this
is what I got:

jabartik extendedTest # python2.1 extended.py
Traceback (most recent call last):
  File "extended.py", line 70, in ?
    build(model, define_properties=1)
  File "/usr/lib/python2.1/site-packages/Modeling/dynamic.py", line 168, in
build
    add_properties(c, e)
  File "/usr/lib/python2.1/site-packages/Modeling/dynamic.py", line 148, in
add_properties
    prop=property(getattr(aClass, 'get'+part_func_name),
NameError: global name 'property' is not defined

I also tried loading (and tweaking for the added Publisher table) the
AuthorBooks example into the Zope ZModeler tool and then using it generate
the code skeletons and database schema; that worked just fine.  I then tried
using these along with the code from Sebastien's example with the via the
python interpreter, which had some issues, as follows:

>>> from Modeling.EditingContext import EditingContext
>>> ec=EditingContext()
>>> from AuthorBooks.Publisher import Publisher
>>> from AuthorBooks.Writer import Writer
>>> from AuthorBooks.Book import Book
>>> p1=Publisher()
>>> p1.bizname="P1"
>>> ec.insert(p1)
>>> ec.saveChanges() 
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
  File "/usr/lib/python2.1/site-packages/Modeling/EditingContext.py", line
737, in saveChanges
    self.validateChanges()
  File "/usr/lib/python2.1/site-packages/Modeling/EditingContext.py", line
765, in validateChanges
    self.objectForGlobalID(gID).validateForSave()
  File "/usr/lib/python2.1/site-packages/Modeling/CustomObject.py", line
595, in validateForSave
    error.finalize()
  File "/usr/lib/python2.1/site-packages/Modeling/Validation.py", line 145,
in finalize
    if self._dict: raise self
Modeling.Validation.ValidationException: Validation for key
OBJECT_WIDE_VALIDATION failed:
- Validation of object <AuthorBooks.Publisher.Publisher instance at
0x849e07c> as a whole failed
Validation for key bizname failed:
- Key is required but value is void

After some scratching, I realized that the setBizname method in
Publisher.py, below, isn't getting called.

  def setBizname(self, bizname):
    "Change the Publisher / bizname attribute value"
    self.willChange()
    self._bizname  = bizname

Doing a dir(p1) revealed:

>>> dir(p1)
['_CustomObject__editingContext', '_bizname', '_books', '_location',
'_v_classDescription', 'bizname']

So the 'bizname' attribute is getting appended and self._bizname = bizname
isn't getting called, which causes the insert to die since bizname is set to
non-null in the database.  Manually forcing it with a p1._bizname="xyz" and
then doing an insert works fine.

So, the questions:

1) Based on some archived posts, I was under the impression that Modeling
core supports Python 2.1 (I am using 2.1.3 since this is leading up to a
Zope 2.6.4 product); is that still true?

2) Do I need the Modeling.dynamic module?  This appears to require python
2.2, but if the classes are being generated via an XML file or the ZModeler
tool, this shouldn't be required, right?

At any rate, thanks again for your time and help.

-Aaron

Re: [Modeling-users] Couple of questions...

[Mario R. <ma...@ru...>]

Hello!

I had come across these kind of limitations a few months ago... not 
developing at the moment, but I find the issue interesting... so, just 
my 2 centimes ;)

> This makes me think that it could be nice to have something like
>
>>>> ec.fetch('Writer', 'NOT EXISTS(books.publisher.bizName!="P1")'

I would say yes this would be very nice to have... and this almost 
seems to me a consistent and natural extension of what is supported 
already, i.e. 'AND', 'OR', 'NOT', 'IN', 'NOT IN', 'like', ilike', ... 
anything else?

Almost... because EXISTS applies to a independent subquery, while the 
other keywords are constraints on the current query. Would this be a 
possible source of confusion?

What happens if we intermix these keywords, e.g.

ec.fetch('Writer', 'books.publisher.numPages>50
AND (books.publisher.bizName IN ["P1","P2"])
AND NOT EXISTS(books.publisher.bizName NOT IN ["P1","p2"] ) '

to get the authors that have books with at least 51 pages as well as 
having published with both of P1 and P2 but with no one else. Would 
this work? There would be a simpler way to do this?

mario

Re: [Modeling-users] Couple of questions...

[Sebastien B. <sbi...@us...>]

John Lenton <jo...@vi...> writes:
> On Tue, Apr 20, 2004 at 09:46:39AM -0700, Aaron Freeman wrote:
[...]
> > How would I get writers by publisher?
> >
> > More specifically, is there an elegant way to do a nested join in a sin=
gle
> > query, or do I just have to do a single writer-book join and then itera=
te
> > over the resultset for the second join?
>=20
> I don't think so. I'd do it with
>=20
>    [ i.valueForKeyPath('books.author') for i in ec.fetch('Publisher') ]
>=20
> which is clean enough for me. I guess what you'd want is for
> valueForKeyPath to be callable on a class... :)


  Yes, that's another approach indeed. But while I'm a fervent defender
for KeyValueCoding ;)) in such cases one should be careful. Here are
some thoughts:

  - if the database have a lot of books, the KVC approach will load them
    all in memory: KVC traverses relationships at the object level, and
    in this case this means fetching Books published by every Publisher,
    then these Books' Authors. This is probably the worst case one can
    get. That's probably why you were asking for a callable vFKP on a
    class, weren't you?

  - In the resulting lists, if an author has published N books for a
    publisher, the corresponding list will receive N duplicates for that
    author.

The one posted earlier:
  >>> ec.fetch('Writer', 'books.publisher.bizName=3D=3D"P1"')
does not have these possible drawbacks.


However and admittedly the dotted notation as it is designed for now can
only handle one kind of semantics for to-many relationships: for
example, there is no way to ask for authors whose books were published
by one single publisher only.

  In other words, then you'll have to use KVC or plain python, traverse
  the relationships and check properties, or, hmmm, get back to plain
  SQL w/ things like:

    SELECT DISTINCT t0.ID, t0.LAST_NAME, t0.FIRST_NAME, t0.AGE,
                    t0.FK_WRITER_ID, t0.BIRTHDAY
    FROM WRITER t0
    WHERE NOT EXISTS (SELECT * FROM BOOK,WRITER,PUBLISHER
                      WHERE WRITER.ID=3Dt0.ID
                            AND WRITER.ID=3DBOOK.FK_WRITER_ID=20
                            AND BOOK.FK_PUBLISHER=3DPUBLISHER.ID
                            AND PUBLISHER.BIZ_NAME!=3D'P1')

  just to get the authors who have no other publisher than P1.


This makes me think that it could be nice to have something like

  >>> ec.fetch('Writer', 'NOT EXISTS(books.publisher.bizName!=3D"P1")'

so that we could avoid having to write plain SQL just like above (yerk:)


        What do you all think?

-- S=E9bastien.

Re: [Modeling-users] Couple of questions...

[John L. <jo...@vi...>]

Sender: John Lenton <jo...@ma...>

On Tue, Apr 20, 2004 at 09:46:39AM -0700, Aaron Freeman wrote:
> (1) Is there a debug flag somewhere that dump all SQL queries to the
> console?
>=20
> (2) How are nested joins accomplished?
>=20
> For example, in the AuthorBooks schema, lets assume there is another table
> called publisher that has a 1-M relationship with books, like so (sorry
> about the formatting):
>=20
> .........................................................................=
...
> .............
> .+-----------+.(0,1)..............(0,*).+-------+.(0,*)..............(0,1=
).+
> .|..Writer...|<-author----------books->>|.Book..|<<-books-------publisher=
->|
> .Publisher.|.
> .|-----------|.(nullify)......(cascade).|-------|........................=
..|
> .|.lastName..|..........................|.title.|........................=
..|
> .bizName...|.
> .|.firstName.|..........................|.price.|........................=
..|
> .location..|.
> .|.age.......|.(0,1)....................+-------+........................=
..+
> .|.birthday..|<-pygmalion--+.............................................=
...
> .............
> .+-----------+.(nullify)...|.............................................=
...
> .............
> =20
> .....|....................|..............................................=
...
> ............
> ......+--------------------+.............................................=
...
> .............
> .........................................................................=
...
> .............
>=20
> How would I get writers by publisher?
>
> More specifically, is there an elegant way to do a nested join in a single
> query, or do I just have to do a single writer-book join and then iterate
> over the resultset for the second join?

I don't think so. I'd do it with

   [ i.valueForKeyPath('books.author') for i in ec.fetch('Publisher') ]

which is clean enough for me. I guess what you'd want is for
valueForKeyPath to be callable on a class... :)

--=20
John Lenton (jo...@vi...) -- Random fortune:
Sobre toda cosa guardada guarda tu coraz=F3n, porque de =E9l emana la vida.
		-- Salom=F3n.=20

Re: [Modeling-users] Couple of questions...

[Sebastien B. <sbi...@us...>]

        Hi Aaron,

"Aaron Freeman" <afr...@em...> wrote:
> (1) Is there a debug flag somewhere that dump all SQL queries to the
> console?

Absolutely: set the environment variable MDL_ENABLE_DATABASE_LOGGING to
any true value. For details, see:
  http://modeling.sourceforge.net/UserGuide/env-vars-core.html


> (2) How are nested joins accomplished?
> For example, in the AuthorBooks schema, lets assume there is another table
> called publisher that has a 1-M relationship with books, like so (sorry
> about the formatting):
>=20
[...
  Writer <-author----books->> Book <<-books----publisher-> Publisher
 ...]

> How would I get writers by publisher?
>=20
> More specifically, is there an elegant way to do a nested join in a single
> query, or do I just have to do a single writer-book join and then iterate
> over the resultset for the second join?
>=20

Yes, there is a dedicated way of doing this: use the dotted notation:

  >>> ec.fetch('Writer', 'books.publisher.bizName=3D=3D"P1"')

returns every writer who published (at least) a book w/ publisher P1.

The generated SQL query is (postgresql here):

  SELECT DISTINCT t0.ID, t0.LAST_NAME, t0.FIRST_NAME, t0.AGE,
         t0.FK_WRITER_ID, t0.BIRTHDAY=20
  FROM WRITER t0=20
  INNER JOIN  ( BOOK t1
                INNER JOIN PUBLISHER t2=20
                ON t1.FK_PUBLISHER=3Dt2.ID )
  ON t0.ID=3Dt1.FK_WRITER_ID
  WHERE t2.BIZ_NAME =3D 'P1';


> Code examples would be great!

  I'm including in the end of the message a PyModel named EAuthorBooks
  (E stands for Extended :) derived from the one used in the User's
  Guide, with sample data, demonstrating the approach.

        Regards,

=20=20=20=20=20=20=20=20
-- S=E9bastien.



PS: you'll need to setup a database before running the example, e.g. w/
    postgresql:

   $ mdl_generate_DB_schema.py \
        --admin-dsn "localhost:template1:postgres:" \
        -v -C pymodel_extAutborBooks.py


------------------------------------------------------------------------
#! /usr/bin/env python
# -*- coding: iso-8859-1 -*-
"Extended AuthorBooks model, with a Publisher"

from Modeling.PyModel import *

##
# Defaults
AFloat.defaults['precision'] =3D 10
AFloat.defaults['scale'] =3D 2
AString.defaults['width'] =3D 40

Association.defaults['delete']=3D['nullify', 'nullify']

Entity.defaults['properties'] =3D [
  APrimaryKey('id', isClassProperty=3D0, isRequired=3D1, doc=3D'PK')
]

##
# Adapt to fit your own configuration!!

_connDict =3D {'database': 'EAUTHOR_BOOKS', 'host':'localhost',
             'user':'postgres','password':''}

model =3D Model('EAuthorBooks',adaptorName=3D'Postgresql',
                connDict=3D_connDict)

model.doc =3D ' ... '
model.version=3D'0.1'
model.entities =3D [
  #
  Entity('Book',
         properties=3D[ AString('title', isRequired=3D1, columnName=3D'titl=
e'),
                      AFloat('price'),
                       ],
         ),
  Entity('Writer',
         properties=3D[ AString('lastName',isRequired=3D1, width=3D30 ),
                      AString('firstName'),
                      AInteger('age', displayLabel=3D'Age'),
                      ADateTime('birthday', usedForLocking=3D0),
                      ]
         ),
  Entity('Publisher',
         properties=3D[ AString('bizName',isRequired=3D1),
                      AString('location'),
                      ]
         ),
  ]
#---
model.associations=3D[
  Association('Book', 'Writer',
              relations=3D['author', 'books'],
              delete=3D['nullify', 'cascade'],
              keys=3D['FK_Writer_Id', 'id']),
  Association('Writer', 'Writer',
              relations=3D['pygmalion', None],
              delete=3D['nullify', None],
              keys=3D['FK_Writer_id', 'id']),
  Association('Book', 'Publisher', relations=3D['publisher', 'books'])
  ]

model.build()
model=3Dmodel.component

if __name__=3D=3D"__main__":
  #import pdb ; pdb.set_trace()
=20=20
  from Modeling.dynamic import build
  build(model, define_properties=3D1)

  from Modeling.ModelSet import defaultModelSet
  defaultModelSet().addModel(model)

  from Modeling.EditingContext import EditingContext
  ec=3DEditingContext()

  from EAuthorBooks.Publisher import Publisher
  from EAuthorBooks.Writer import Writer
  from EAuthorBooks.Book import Book

  # Initialization
  p1=3DPublisher(); p1.bizName=3D"P1"
  p2=3DPublisher(); p2.bizName=3D"P2"

  w1=3DWriter(); w1.lastName=3D"in p1"
  w2=3DWriter(); w2.lastName=3D"in p2"
  w3=3DWriter(); w3.lastName=3D"in p1 and p2"
=20=20
  b1w1=3DBook(); b1w1.title=3D"b1w1"
  b2w1=3DBook(); b2w1.title=3D"b2w1"
  b1w2=3DBook(); b1w2.title=3D"b1w2"
  b2w2=3DBook(); b2w2.title=3D"b2w2"
  b1w3=3DBook(); b1w3.title=3D"b1w3"
  b2w3=3DBook(); b2w3.title=3D"b2w3"
  ec.insert(p1);
  ec.insert(p2); ec.insert(w1); ec.insert(w2); ec.insert(w3)
  ec.insert(b1w1); ec.insert(b2w1); ec.insert(b1w2); ec.insert(b2w2)
  ec.insert(b1w3); ec.insert(b2w3)

  w1.addToBooks(b1w1); b1w1.author=3Dw1; w1.addToBooks(b2w1); b2w1.author=
=3Dw1
  w2.addToBooks(b1w2); b1w2.author=3Dw2; w2.addToBooks(b2w2); b2w2.author=
=3Dw2
  w3.addToBooks(b1w3); b1w3.author=3Dw3; w3.addToBooks(b2w3); b2w3.author=
=3Dw3

  b1w1.publisher=3Dp1; p1.addToBooks(b1w1)
  b2w1.publisher=3Dp1; p1.addToBooks(b2w1);
  b1w2.publisher=3Dp2; p2.addToBooks(b1w2)
  b2w2.publisher=3Dp2; p2.addToBooks(b2w2);
  b1w3.publisher=3Dp1; p1.addToBooks(b1w3)
  b2w3.publisher=3Dp2; p2.addToBooks(b2w3);

  ec.saveChanges()

  # query
  ec=3DEditingContext()
  print [w.lastName
         for w in ec.fetch('Writer', 'books.publisher.bizName=3D=3D"P1"')]
  # returns: w1 and w3
  ec=3DEditingContext()
  print [w.lastName
         for w in ec.fetch('Writer', 'books.publisher.bizName=3D=3D"P2"')]
  # returns: w2 and w3
------------------------------------------------------------------------

[Modeling-users] Couple of questions...

[Aaron F. <afr...@em...>]

(1) Is there a debug flag somewhere that dump all SQL queries to the
console?

(2) How are nested joins accomplished?

For example, in the AuthorBooks schema, lets assume there is another table
called publisher that has a 1-M relationship with books, like so (sorry
about the formatting):

............................................................................
.............
.+-----------+.(0,1)..............(0,*).+-------+.(0,*)..............(0,1).+
-----------+.
.|..Writer...|<-author----------books->>|.Book..|<<-books-------publisher->|
.Publisher.|.
.|-----------|.(nullify)......(cascade).|-------|..........................|
-----------|.
.|.lastName..|..........................|.title.|..........................|
.bizName...|.
.|.firstName.|..........................|.price.|..........................|
.location..|.
.|.age.......|.(0,1)....................+-------+..........................+
-----------+.
.|.birthday..|<-pygmalion--+................................................
.............
.+-----------+.(nullify)...|................................................
.............
 
.....|....................|.................................................
............
......+--------------------+................................................
.............
............................................................................
.............

How would I get writers by publisher?

More specifically, is there an elegant way to do a nested join in a single
query, or do I just have to do a single writer-book join and then iterate
over the resultset for the second join?

Code examples would be great!

Thanks for your time.

-Aaron

Re: [Modeling-users] problem with qualifier parser?

[Sebastien B. <sbi...@us...>]

        Hi John and all,

  You'll find a patch solving the problem along w/ the corresponding
  unittests at sf.net, bug #938096:

https://sourceforge.net/tracker/index.php?func=3Ddetail&aid=3D938096&group_=
id=3D58935&atid=3D489335

        Thanks for reporting!

-- S=E9bastien.


PS for Yannick: I still did not find the time to investigate any further
  about GNU Arch, but I will, this looks interesting (not to say about
  the new name for the framework...!-). Any comparison w/ subversion,
  BTW?


> John Lenton <jo...@vi...> writes:
> >     $ python
> >     Python 2.3.3 (#2, Feb 24 2004, 09:29:20)
> >     [GCC 3.3.3 (Debian)] on linux2
> >     Type "help", "copyright", "credits" or "license" for more informati=
on.
> >     >>> from Modeling.EditingContext import EditingContext
> >     >>> ec=3DEditingContext()
> >     >>> ec.fetch('foo', 'orbit > 3')
> >     Traceback (most recent call last):
> >       File "<stdin>", line 1, in ?
> [...]
> >         raise ValueError, "Syntax error near token: `%s'" % token
> >     ValueError: Syntax error near token: `<Token OR>'
> >=20
> > actually there's the same problem with any token: OR, AND, IN, or
> > NOT. Is this a known problem in spark? is my spark version wrong?
> > spark is 0.6.1, Modeling is 0.9-pre-16-patch892454.
>=20
>=20
> This is a bug, unfortunately not fixed on sf yet :/  I'll have a look at =
it
> tonite. In the meantime you can build the same qualifier by hand:
>=20
>   >>> from Modeling.Qualifier import *
>   >>> q=3DKeyValueQualifier('foo', QualifierOperatorGreaterThan, 3)
>   >>> print q
>   (foo>3)
>   >>> ec.fetch('foo', q)
>=20
> You can have a look at test_Qualifier.py for other examples.

Re: [Modeling-users] Python 2.3 and booleans

[Yannick G. <ygi...@yg...>]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On April 16, 2004 08:25 am, Sebastien Bigaret wrote:
> =A0 =A0 def _setMyAttribute(self, value):
> =A0 =A0 =A0 =A0 self.value=3Dbool(value)
>
> (this is in fact the very same technique exposed here:
> =A0http://modeling.sourceforge.net/UserGuide/attribute-custom-type.html)
>
>
> =A0 =A0 If it's not what you need, could you also give a little example?

Thats exactly what I need (except that I want an int instead of a bool) !

Thanks !

 : )

> =A0 and I'll have a look at Arch when I find time ;) =A0Any idea on wheth=
er
> =A0 it wirks on sf?

What is nice about Arch is that is does not need any support on the
server side.  Any remote file storage method will work, FTP, SFTP,
WebDav, NFS...

There is a ViewArch script to give the same functionality as ViewCVS.
You can also have free hosting on sourcecontrol.net as long as your
archive contain only free software.  Sourcecontrol.net offers some
more Arch centric support like remote branch searching and they are in
the process to add website and release hosting too.

The Emacs vc-arch mode also made it's way in the Emacs CVS so we can
expect to see it in the next release.

Compared to Subversion, the other likely replacement for CVS, Arch
standout because it does not require special action on the server, it
already support symlinks and you have GPG signature of commits which
is quite an important feature with all the recent development server
compromises.  Both support renaming and moving but Arch can detect it
without notifications without breaking history if you use tag-lines.
Thats quite a nice feature if you come up with the new name one of
these days...

 ; )

Get more information here:

  http://wiki.gnuarch.org/

=2D --=20
Yannick Gingras     "Let's say the docs present a simplified view of realit=
y"
Coder for OBB : Obliquely Berserk Bookstore                     -- Larry Wa=
ll
http://OpenBeatBox.org

=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFAgA0jZJ8/OobqizMRAoacAJ9TmW9EFYfoIpwW6S9Xth+3jmm8WACgg7ZP
9W8GP89mmfusGsJhpL2jXUA=3D
=3DchOf
=2D----END PGP SIGNATURE-----

Re: [Modeling-users] Python 2.3 and booleans

[Sebastien B. <sbi...@us...>]

        Hi Yannick,

  Not sure I fully understand what effect you want here, just in case
this can help: you can transform any attribute's value stored as an int
in the db into py2.3 boolean automaticall=3Dy by using a KeyValueCoding
private setter, like that:


    def _setMyAttribute(self, value):
        self.value=3Dbool(value)

(this is in fact the very same technique exposed here:
 http://modeling.sourceforge.net/UserGuide/attribute-custom-type.html)


    If it's not what you need, could you also give a little example?

(admittedly I've read your msg quickly and answering the same way...)

  and I'll have a look at Arch when I find time ;)  Any idea on whether
  it wirks on sf?

-- S=E9bastien.



Yannick Gingras <ygi...@yg...> writes:

> Hi,=20
>   we just upgraded to Python 2.3.  As you probably know, Python 2.3
> now has a boolean type so repr(a =3D=3D b) is no longer 0 or 1 but True or
> False.
>=20
> We use some int(1) in our database to represent booleans but now all
> our queries involving those fields are broken.
>=20
> This is absolutely not a Modeling problem but I think that I can use
> Modeling to reconvert those values just before the query.  That would
> solve quite a lot of work here.
>=20
> Any of you guys had the same problem when switching to Python 2.3 ?
>=20
> Should I tweak Modeling to make the conversion automatically ?  Can it
> break anything somewhere else ?
>=20
> And absolutely unrelated to this problem, I've been using GNU Arch
> lately instead of CVS.  It features GPG signed commits and ease
> maintaining branches in synch.  If Modeling used Arch, we could make
> the changes to automate booleans conversions in our branch and still
> stay up to date with the main branch in case someone else would like
> to have this particular feature and Sebastian find our code too ugly
> to be merged yet...
>=20
>  ; )
>=20
> Enough propaganda, thanks for your time guys !
>=20

Re: [Modeling-users] problem with qualifier parser?

[Sebastien B. <sbi...@us...>]

        Hi John,

John Lenton <jo...@vi...> writes:
> Hi there, long time no see :)
>=20
> First, sorry for my delay with the cimarron example. I've had to stop
> working on cimarron for a while...

Not a problem, 'having a hard time here as well :)

> Now I'm working with Modeling in zope, and I seem to have come across
> a bug... and I'm hoping there's an easy fix (or that it's already
> fixed and I've just not seen it on sourceforge :p).
>=20
>     $ python
>     Python 2.3.3 (#2, Feb 24 2004, 09:29:20)
>     [GCC 3.3.3 (Debian)] on linux2
>     Type "help", "copyright", "credits" or "license" for more information.
>     >>> from Modeling.EditingContext import EditingContext
>     >>> ec=3DEditingContext()
>     >>> ec.fetch('foo', 'orbit > 3')
>     Traceback (most recent call last):
>       File "<stdin>", line 1, in ?
[...]
>         raise ValueError, "Syntax error near token: `%s'" % token
>     ValueError: Syntax error near token: `<Token OR>'
>=20
> actually there's the same problem with any token: OR, AND, IN, or
> NOT. Is this a known problem in spark? is my spark version wrong?
> spark is 0.6.1, Modeling is 0.9-pre-16-patch892454.


This is a bug, unfortunately not fixed on sf yet :/  I'll have a look at it
tonite. In the meantime you can build the same qualifier by hand:

  >>> from Modeling.Qualifier import *
  >>> q=3DKeyValueQualifier('foo', QualifierOperatorGreaterThan, 3)
  >>> print q
  (foo>3)
  >>> ec.fetch('foo', q)

You can have a look at test_Qualifier.py for other examples.

        HTH,

-- S=E9bastien.

[Modeling-users] Python 2.3 and booleans

[Yannick G. <ygi...@yg...>]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Hi,=20
  we just upgraded to Python 2.3.  As you probably know, Python 2.3
now has a boolean type so repr(a =3D=3D b) is no longer 0 or 1 but True or
=46alse.

We use some int(1) in our database to represent booleans but now all
our queries involving those fields are broken.

This is absolutely not a Modeling problem but I think that I can use
Modeling to reconvert those values just before the query.  That would
solve quite a lot of work here.

Any of you guys had the same problem when switching to Python 2.3 ?

Should I tweak Modeling to make the conversion automatically ?  Can it
break anything somewhere else ?

And absolutely unrelated to this problem, I've been using GNU Arch
lately instead of CVS.  It features GPG signed commits and ease
maintaining branches in synch.  If Modeling used Arch, we could make
the changes to automate booleans conversions in our branch and still
stay up to date with the main branch in case someone else would like
to have this particular feature and Sebastian find our code too ugly
to be merged yet...

 ; )

Enough propaganda, thanks for your time guys !

=2D --=20
Yannick Gingras                                "please call it ``GNU/Linux'=
'"
Coder for OBB : Officious Billowing Bailey             -- Richard M. Stallm=
an
http://OpenBeatBox.org

=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFAfyG1ZJ8/OobqizMRAvk8AKDojHEvO8hdhLmk1xMtKfZMn4aKxwCeIhq9
quoHiRK/W5PBstlw9qs+DOE=3D
=3Dk51v
=2D----END PGP SIGNATURE-----

[Modeling-users] problem with qualifier parser?

[John L. <jo...@vi...>]

Sender: John Lenton <jo...@ma...>

Hi there, long time no see :)

First, sorry for my delay with the cimarron example. I've had to stop
working on cimarron for a while...

Now I'm working with Modeling in zope, and I seem to have come across
a bug... and I'm hoping there's an easy fix (or that it's already
fixed and I've just not seen it on sourceforge :p).

    $ python
    Python 2.3.3 (#2, Feb 24 2004, 09:29:20)=20
    [GCC 3.3.3 (Debian)] on linux2
    Type "help", "copyright", "credits" or "license" for more information.
    >>> from Modeling.EditingContext import EditingContext
    >>> ec=3DEditingContext()
    >>> ec.fetch('foo', 'orbit > 3')
    Traceback (most recent call last):
      File "<stdin>", line 1, in ?
      File "/usr/lib/python2.3/site-packages/Modeling/EditingContext.py", l=
ine 1413, in fetch
        qualifier=3DqualifierWithQualifierFormat(qualifier)
      File "/usr/lib/python2.3/site-packages/Modeling/Qualifier.py", line 1=
29, in qualifierWithQualifierFormat
        return QualifierParser.qualifierWithQualifierFormat(expression)
      File "/usr/lib/python2.3/site-packages/Modeling/QualifierParser.py", =
line 550, in qualifierWithQualifierFormat
        _parse=3Dparse(sc)
      File "/usr/lib/python2.3/site-packages/Modeling/QualifierParser.py", =
line 417, in parse
        return parser.parse(tokens)
      File "/usr/lib/python2.3/site-packages/spark.py", line 204, in parse
        self.error(tokens[i-1])
      File "/usr/lib/python2.3/site-packages/Modeling/QualifierParser.py", =
line 261, in error
        raise ValueError, "Syntax error near token: `%s'" % token
    ValueError: Syntax error near token: `<Token OR>'

actually there's the same problem with any token: OR, AND, IN, or
NOT. Is this a known problem in spark? is my spark version wrong?
spark is 0.6.1, Modeling is 0.9-pre-16-patch892454.

--=20
John Lenton (jo...@vi...) -- Random fortune:
Teme a la vejez, pues nunca viene sola.
		-- Plat=F3n. (427-347 a.C.) Fil=F3sofo griego.=20

Re: [Modeling-users] What for?

[Sebastien B. <sbi...@us...>]

        Hi,

  ...And sorry for the late answer

Luciano Andrade <and...@ya...> wrote:
> I whant to know what meens the attributeUseForLooking,

The 'usedForLocking' property is not used yet --it will be used to
indicate which attributes should be monitored when the forthcoming
optimistic locking feature is implemented.

>  what mees that a relation is a classpropety, what if
> a propiety is not a classpropety,=20

The 'classProperty' flag indicates whether the related attribute is part
of the class' attributes. For example, 'firstName' is a class property
in entity Person: any object of class Person has a property 'firstName';
on the contrary, the primary key is not a class property, it is just a
column in the database that is not exposed at the object level (so in
this case the DB's table has a 'id' column but class Person does not
have any 'id' attribute)

> what the join's semantics meens,

This has to do with different sets of objects returned by a SQL query
--I myself only use inner joins (ie themost natural one, at least for
me;), so instead of trying to explain things I do not master at all, I
prefer to give you some links that ill do the job better then I could:


  Getting the Right Data with SQL Joins:
    http://www.devx.com/dbzone/Article/17403/0/page/1
=20=20
  Fropm postgresql:
  http://www.postgresql.org/docs/7.4/interactive/tutorial-join.html
=20=20
  O'Reilly:
    (pointed to by: http://www.onlamp.com/pub/ct/19)
=20=20
    http://www.onlamp.com/pub/a/onlamp/2001/05/24/aboutSQL.html
    http://www.onlamp.com/pub/a/onlamp/2001/06/01/aboutSQL.html
    http://www.onlamp.com/pub/a/onlamp/2001/06/15/aboutSQL.html
    http://www.onlamp.com/pub/a/onlamp/2001/06/27/aboutSQL.html

> what about circular relations, can the module handled.


  Not sure what you mean: do you mean reflexive relationships,
  i.e. where the source and destination entities are the same entity?
  If this is the case, have a look at the sample model in the User's
  Guide:

     http://modeling.sourceforge.net/UserGuide/model-author-books.html

  It has a reflexive relationship Writer-->Writer (it is a directional
  relationship, but you can have a one-to-many bi-directional
  relationship as easily), which is modeled exactly the same than
  "standard" relationships. The corresponding PyModel is exposed here:

     http://modeling.sourceforge.net/UserGuide/pymodel-sample.html


Should you have other questions, feel free to ask for more here!


-- S=E9bastien.

[Modeling-users] What for?

[<and...@ya...>]

I whant to know what meens the attributeUseForLooking,
 what mees that a relation is a classpropety, what if
a propiety is not a classpropety, what the join's
semantics meens, what about circular relations, can
the module handled.

------------
Los mejores usados y las más tentadoras 
ofertas de 0km están en Yahoo! Autos.
Comprá o vendé tu auto en
http://autos.yahoo.com.ar

Re: [Modeling-users] dirty deeds done dirt cheap

[Sebastien B. <sbi...@us...>]

Marcos Dione <md...@vi...> wrote:
> On Thu, Mar 25, 2004 at 01:03:00AM +0100, Sebastien Bigaret wrote:
> > Repopulating the EC simply consists in fetching the objects, or asking
> > for a particular object given its globalID (ec.faultForGlobalID()), etc.
>=20
>     well, I was thinking in that yesterday when I left work... is it too
> ugly to do it that way? will it mess with something?

No it's not: GlobalIDs are especially designed to designate a specific
object/row, and using a child EC has several advantages. I was just
saying then that the features it provides [mostly: "OO-transactions" and
inheriting the changes made in the parent EC(s)] may be oversized if all
you need is to forget the state of one or a few objects.

And if you mess with something when doing it this way, that means you've
probably been bitten by a bug.

>
[...]
> --- paste ---
> I've quickly checked the implementation & at the test I proposed there,
> so here is my suggestion: you can use it as-is, given that:
>=20=20=20=20=20
> - you're not applying it on deleted, inserted or modified objects, (BTW:
>   it will become an error to refault a deleted or an inserted object)
>   [UPDATE: see note, below]
>=20
> - you stick to EC.refaultObject and do not try to call
>   DBContext.refaultObject
>=20=09=20=20=20
> - you do not use it with a nested ec.
>=20=09=20=20=20=20
> (I'm not saying this won't work, it probably will... but it needs to be
> tested before I can say this is supported).
> --- paste ---
>=20
>     that last remark applies to the last contition? from the patch, it
> seems like the refaultObject is finally done by the 'master' of all the
> EC's. do you think the problem would be that the 'reset' is not
> 'propagated' to children EC's?

- Exactly, it is indeed propagated down the EC hierarchy, then to the
  ObjectStoreCoordinator and ultimately to the DatabaseContext
  responsible for the object's entity.

- The last remark was targetted to the three conditions. You may
  wonder why making this available to non-modified, permanent objects ->
  reason is that it was initially designed to make it possible to
  "release" objects in ECs, e.g. to keep the number of held objects
  under a certain limit: releasing an object also release memory (from
  db caches, mostly).

    However it's applicable to modified objects, as shown in the test.

- About the propagation of reset up the EC hierarchy: it's not done and
  this is a problem. Most probably, the rest of an object wil also reset
  it in its parents. But even then, this is only a proposal for the
  default behaviour, and one ec should be able to provide its own
  response to the refault notification coming from a parent, for example
  through a delegate.


NB: however to test it w/ modified objects the following lines in
    EC.refaultObject() should be disabled --my fault:

        if gid in self._pendingUpdatedObjects+self._updatedObjects:
          raise ValueError, 'Cannot refault a modified object'

>=20=20=20=20=20
>     the other comment is (I think this is the update you mention above):
>=20=20=20=20=20
> --- paste ---
> Note: it can be safely applied on modified objects given that no
> relationships have been modified (such a situation has not been tested
> at all).
> --- paste ---
>=20
>     I'll try to make some tests in both cases.
>=20=20=20=20=20
>     it is interesting to note that most of this patch is already in the
> slice_n_sort patch. is that an error? shouldn't this other one depend on
> the first one, so applying both doesn't choke?

You're right!  I was not even able to find a difference, it's completely
included in patch #892454... I have several tests configurations here
and it seems that one got integrated in the other. Thanks for noticing,
I'll add a note to the patches'page.

-- S=E9bastien.