Recent changes to support-requests

The speed of SQLObject is slower on the time

Rosa Robles — Thu, 14 Jul 2011 07:40:43 -0000

I had a test of a SQLObject and I had found a strange behaviour of SQLObject. On the time, the respond time is lesser.

Here, I put a example of test code :

from sqlobject.sqlbuilder import *
connection = connectionForURI('postgresql://test:test@localhost/test_sql_object')
sqlhub.processConnection = connection

class Person(SQLObject):
firstName = StringCol()
middleInitial = StringCol(length=1, default=None)
lastName = StringCol()

iteration = 0
while 1:
start = datetime.now()
person_list = []
for key in range(1000):
person = Person.selectBy(firstName='name%s%s' % (iteration,key))[0]
person_list.append(person)

for person in person_list:
person.expire()
del person
del person_list
sqlhub.processConnection.cache.clear()

end = datetime.now()

print "TIME ",iteration,':' , end - start
iteration = iteration + 1

In the result I can check that in each iteration the time is bigger. The first iteration has time of 1.37s, then the time increases and the 19 iteration has time of 5.83 s!

TIME 0 : 0:00:01.376392
TIME 1 : 0:00:01.663250
TIME 2 : 0:00:01.978639
TIME 3 : 0:00:02.319939
TIME 4 : 0:00:02.618213
TIME 5 : 0:00:02.723846
TIME 6 : 0:00:02.584007
TIME 7 : 0:00:02.868877
TIME 8 : 0:00:03.051937
TIME 9 : 0:00:03.389757
TIME 10 : 0:00:03.503036
TIME 11 : 0:00:03.563154
TIME 12 : 0:00:04.191539
TIME 13 : 0:00:04.274464
TIME 14 : 0:00:04.813206
TIME 15 : 0:00:04.677002
TIME 16 : 0:00:04.995708
TIME 17 : 0:00:05.126811
TIME 18 : 0:00:05.571279
TIME 19 : 0:00:05.832840

Why is happend this?
Thanks in advance for your attention.
Cheers Rosa

SQLObject documentation

Michel Albert — Wed, 16 Aug 2006 07:55:56 -0000

Hi,

SQLObject is a brilliant project and it's much fun
working with it..... Once you figured out the details.
SQLObject suffers from a *horrid* documentation.

Starting a project from scratch is explained well
enough. But if you have to re-use an existing
database. It's inevitable that you need to use more
advanced features of SQLObject. And those are only
badly (if at all) documented. For example:

You can use a different "Style" object to indicate
a different naming convention.

But which "Style" objects exist? What are the
parameters? Is it possible to create a Style object
yourself? If so, how? What methods need to be
overridden and what do they need to do?

You might say: "Hey, it's open-source. Browse the
source code and see for yourself." And that's fine by
me. But then, *please* document the code! My first try
was to open up the python interpreter and do a
>>> help(SQLObject)
and go from there. Nothing useful showed up.

Honestly, for me time is money, and I was really close
to giving up with SQLObject for this reason. However I
like SQLObject. So I'll take a sip of vinegar and
stick with it. After doing a lot of digging, I got
most things to work. But not necessarily the way I
want them to. Alas, it works.

The gist of this message is: There might be people
less willing to do some research, and drop SQLObject
quickly. It would be sad, if SQLObject suffers from
it's lack of documentation. And if I understood more
of it's inner-workings, I would offer you to write
some documentation, and even redesign the web-page, as
-- quite frankly -- it has a bit of a chaotic look.
The way I understand it, it's automatically generated,
so it's not your fault.

Anyhow, I hope you read this message as positive
criticism and not as a rant/flame. As it was not
intended as such.

Best regards,

Michel Albert

p.s.: is it possible to do a naming style that works
like mixed-case, but keeps the first letter small? Or
even better, one that completely keeps the style of
the variable (does no modifications whatsoever)?

SQLite table name mismatch

Anonymous — Thu, 29 Sep 2005 19:37:11 -0000

Hello,

I am fairly new to python and SQLObject. I have been
asked to resurrect some code and have been having a
reoccurring problem: as near as I can tell, the column
name to object name translation intermittently
mismatches (omits/adds a _ in middle of name before a
capital.)

EG
For relationItem in table.relationThing:
Print relationItem
Will consistently generate a database error no such
column: relation_thing_id . The column is named
relationthing_id. The tables are generated form the
classes, and have a consistent naming schema for
foreign keys. All other times this relation is
referenced, it generates the right table name.

I have spent a few weeks trying combinations of
different versions of python (2.3 and 2.4), pysqlite
(2.0.4, 1.1.6, 1.0), sqlite (2.8, 3.2.5) and OS (winXP,
SuSE, Debian) and have not had any success elevating
this problem. The code ran at one point, but the box
that it was running on has since been formatted. I
have also tried the SQLObject version 7 with same
results. These configurations pass the regression testing.

This looks to be an amazing tool, and I would like to
include it in future projects, provided that I can get
it working.

Any help would be welcome,
matt inthefrist@hotmail.com

ForeignKey Column Name Limitation

aledain — Thu, 22 Sep 2005 04:03:51 -0000

This is probably a feature not a bug. It might be how
SQLObject does some of its magic. If I do the following:

class Person (SQLObject) :
dbName = StringCol()
dbAddresses = MultipleJoin('Address')

class Address (SQLObject) :
dbStreet = StringCol()
person = ForeignKey('Person')

# Assume tables are created.

p = Person(dbName='me')
print p.dbAddresses

a = Address(dbStreet='street',person=p)
print p.dbAddresses
print a.person

This works as expected. However, if I change 'person'
in the Address class to 'dbPerson' I get a traceback
ending with: DatabaseError: no such column: person_id.
This traceback occurs at line p = Person(dbName='me').

It appears that the column name in Addresses,
ForeignKey'ed to Person must match the class name. In
addition, you cannot overwrite the column name using
dbPerson = ForeignKey('Person', dbName='person').

Am I correct that the column name must match the
reference class to use ForeignKey?

cheers, Alex.

total newbie using CherryPy & SQLObject

Maureen Black — Wed, 25 May 2005 00:51:34 -0000

The following code sample seems to only update via
cache, and I don't see the actual record updated nor
inserted in the database when I call 'store'. I turned
on debugging and I can see the SQL statements. I can
attach to MySQL and query it directly via localhost.
select works fine, too. I can see a row I manually
enter via the web form.

Any assistance is appreciated!

----------------

from cherrypy import cpg
from Cheetah.Template import Template
from sqlobject.mysql.mysqlconnection import MySQLConnection

conn = MySQLConnection(user='root', db='test',
passwd="", debug=1)

# this is our (only) data class.
class Contact(SQLObject):
_connection = conn
lastName = StringCol(length = 50, notNone = True)
firstName = StringCol(length = 50, notNone = True)
phone = StringCol(length = 30, notNone = True,
default = '')
email = StringCol(length = 30, notNone = True,
default = '')
url = StringCol(length = 100, notNone = True,
default = '')

class ContactManager:
def index(self):
# Let's display a list of all stored contacts.
contacts = Contact.select()

template = Template('''
<h2>All Contacts</h2>

#for $contact in $contacts
<a
href="mailto:$contact.email">$contact.lastName,
$contact.firstName</a>
[<a href="./edit?id=$contact.id">Edit</a>]
[<a
href="./delete?id=$contact.id">Delete</a>]
<br/>
#end for

<p>[<a href="./edit">Add new contact</a>]</p>
''', [locals(), globals()])

return template.respond()

index.exposed = True

def edit(self, id = 0):
# we really want id as an integer. Since
GET/POST parameters
# are always passed as strings, let's convert it.
id = int(id)

if id > 0:
# if an id is specified, we're editing an
existing contact.
contact = Contact.get(id)
title = "Edit Contact"
else:
# if no id is specified, we're entering a
new contact.
contact = None
title = "New Contact"

# In the following template code, please note
that we use
# Cheetah's $getVar() construct for the form
values. We have
# to do this because contact may be set to None
(see above).
template = Template('''
<h2>$title</h2>

<form action="./store" method="POST">
<input type="hidden" name="id"
value="$id" />
Last Name: <input name="lastName"
value="$getVar('contact.lastName', '')" /><br/>
First Name: <input name="firstName"
value="$getVar('contact.firstName', '')" /><br/>
Phone: <input name="phone"
value="$getVar('contact.phone', '')" /><br/>
Email: <input name="email"
value="$getVar('contact.email', '')" /><br/>
URL: <input name="url"
value="$getVar('contact.url', '')" /><br/>
<input type="submit" value="Store" />
</form>
''', [locals(), globals()])

return template.respond()

edit.exposed = True

def delete(self, id):
# Delete the specified contact
contact = Contact.get(int(id))
contact.destroySelf()
return 'Deleted. <a href="./">Return to Index</a>'

delete.exposed = True

def store(self, lastName, firstName, phone, email,
url, id = None):
if id and int(id) > 0:
# If an id was specified, update an
existing contact.
contact = Contact.get(int(id))

# We could set one field after another, but
that would
# cause multiple UPDATE clauses. So we'll
just do it all
# in a single pass through the set() method.
contact.set(
lastName = lastName",
firstName = firstName,
phone = phone,
email = email,
url = url)
else:
# Otherwise, add a new contact.
contact = Contact(
lastName = lastName,
firstName = firstName,
phone = phone,
email = email,
url = url)

return 'Stored. <a href="./">Return to Index</a>'

store.exposed = True

def reset(self):
# Drop existing table
Contact.dropTable(True)

# Create new table
Contact.createTable()

# Create some sample data
Contact(
firstName = 'testfirstName',
lastName = testLastName',
email = 'test@test.com',
phone = '555-1212',
url = 'http://www.nope.com')

return "reset completed!"

reset.exposed = True

print "If you're running this application for the first
time, please go to http://localhost:8080/reset once in
order to create the database!"

cpg.root = ContactManager()
cpg.server.start(configFile = 'tutorial.conf')

One-to-One Join

Anonymous — Tue, 13 May 2003 21:12:29 -0000

This might be trivial - but can you show an example of
a one-to-one join using the _joins attribute.

Thanks!