Hallo there!
I've began exploring TG2.1 today after years of 1.x, and I like it.
It would be nice to start writing new modules for an existing business
application, currently running under TG1.1.
The current app has been in production since 2007, and porting the
whole baby is unthinkable.
On 01/29/2010 11:23 AM, turbogears+nore...@googlegroups.com wrote:
I *think* this is because you need to have the name SQLAlchemy
defined in the
module you define the handler in. So do you import it in your
controller
file?
This was how it behaved some time
On May 20, 5:47 pm, Matt Wilson m...@tplus1.com wrote:
I like the idea of using a truncate statement with all my tables to
wipe out all the rows and then reloading them. I don't know how to
use a rollback with controller tests because cherrypy (or TG) seems to
issue a commit at the end of
On May 20, 12:03 am, Daniel Fetchinson fetchin...@googlemail.com
wrote:
I don't want to manually wipe out all the entries because it is a
hassle to keep updating my teardown method after each new test. So,
how can I purge all the data quickly?
And how about having a small in-memory
Julian Yap wrote:
Using formencode.validators, I have a field which accepts an Int. I
also want it to validate NotEmpty.
I sugges you to Read The Formencode Manual
Two ways:
from formencode.validators import Int, NotEmpty, All
validator=Int(min=100, max=999, not_empty=True)
seber...@spawar.navy.mil wrote:
TG or Rails or Django have no magic dust, they just make a lot of
assumptions (RoR even more) that bigger frameworks don't.
What are these assumptions that Zope/J2EE/ASP crowd don't make?
Just an example. Coming from zope, I found weird at first
Michael Pearce wrote:
Hi,
I have created a WidgetsList object which contains a number of
attributes (individual widgets) for creating a database record. I want
to use the same WidgetList for editing the database record, however
the edit page needs to display additional fields that are
Sergei Beilin wrote:
1) I wonder if tgext's CrudRestController is close to your specs?
2) I suppose that model/id/action is much better then your's model/
action/id
I suppose the REST people have said something about compound primary
keys, btw? How would you handle them?
ruko wrote:
I thought it is sufficient to have the # -*- coding: utf-8 -*- at the
top of the file. It seems as if some libraries do not expect unicode.
You are mistaking Unicode and UTF-8. A plain string is encoded in UTF8
if you have the # -*- header, but a unicode object is another
edgarsmolow wrote:
It's too bad that ORMs, whether SA or SO, have limited reflection
capabilities. That coerces a one-way methodology: design with the
ORM.
How is SA's reflection limited?
I ask because in my current app (about 150 tables and triggers and
stuff) I have never felt
Jorge Vargas wrote:
For the record that statement was aimed mainly at SO, and it does
coerces a structure on your tables which you need to mimic in order to
get done.
I reckon that. I choose TG over Django for its SQLAlchemy support.
Since mine is a management control application (with
Ben Sizer wrote:
likely to be fairly static. It's not a big deal either way since I
expect the implementation will end up 95% the same whichever route I
take, but I'm just trying to get a feel for what value widgets provide
over something like py:def and what they don't.
Widgets
Victor wrote:
class Controller(tg.Controller):
@tg.expose()
def admin(self, ...):
pass
@tg.expose()
def post(self, ...)
pass
@tg.expose()
def view(self, ...)
pass
I want the first directory name be a parameter. Is that possible to
Victor wrote:
@tg.expose()
def default(self, name, action, *args, **kwargs):
redirect(tg.url('/site', dict(name=name, action=action,
**kwargs)))
Just a note: redirect() is supposed to be raised, so that you can
clearly see there is a disruption of program flow (like with
Marco Mariani wrote:
You can raise cherrypy.InternalRedirect, but it won't change your
request parameters: name and action will still be there.
Sorry, I'm not sure what I wrote about request parameters, I've seldom
used it, try
Hallo there!
I'm trying to port my current 2 yrs old project (running on TG 1.0.8 +
SA 0.4.8 + postgres) to TG 1.1b3
It's a management control application that relies heavily on
SqlAlchemy's autoload feature due
to the complexity of the db (150+ tables and views, growing)
I really think the
Felix Schwarz wrote:
I agree completely with that one. In the end I deployed a solution
which does not accept '.' for German decimal numbers if they are not
used as grouping separator.
Do users really enter grouping separators in program forms?
Have you seen them in the wild? :)
Kirk Strauser wrote:
First, how is that even possible? I've never seen a Python class that
behaves
that way (not meant as a criticism, but genuinely curious).
You should look at the WidgetsList implementation, it's not hard.
Second, are there idiomatic workarounds?
Yes,
Kirk Strauser wrote:
So, I basically need to convince TG and SQLAlchemy that customers.groupcode
is unique so I can display those values in popups and create a many-to-many
table linking users_table to customers.
Don't you have another column you can use with groupcode, to uniquely
Mark Ramm wrote:
And really Genshi is very similar to Kid, so that
transition is pretty easy.
Well... how do I port my TG widgets over to Genshi, in 1.0.5 or 1.5 or
whatever?
Since I cannot mix TG and Tosca widgets on the same form, I am pretty
much forced to port all of the widgets to
[EMAIL PROTECTED] wrote:
But since no one has come up with a good reason for having files with
the same name (not the same namespace, just the same name),
Because sensible, readable names under namespaces are what distinguishes
us from COBOL.
I standy-by my original assertion that it's an
[EMAIL PROTECTED] wrote:
Wait a minute. I've never, ever imported turbogears.controllers.
Really?
Really. Why should I keep dangling imports? validate, expose(),
flash(), redirect(), url() are available through turbogears.__init__
I could swear that the first line of the
Dean Landolt wrote:
Having modules with the same basename in different packages is
general practice in Python.
This sounds like it might be some confusion do to a from turbogears
import *
Not as confusing as from os import *, where your beloved open()
builtin goes down the
Max wrote:
Is there any reason why an Identity session might be reset on a
page refresh?
Check for the session_id and tg-visit cookies: is one of them missing?
Does it only happen on firefox?
If the answer is yes, I've seen it twice. I made it work by using the
'max-age' property
Max wrote:
There is no session_id cookie before or after login.
Which is fine because the stock identity does not use a session.
--
This e-mail (and any attachment(s)) is strictly confidential and for use only
by intended recipient(s). Any use, distribution, reproduction or disclosure by
Diez B. Roggisch wrote:
I see I could add:
@identity.require(identity.in_any_group(admin,group2,group3))
I think you should grant a permission to all of those groups, and check
the permission directly. That's what permissions are for.
Or else, in case things get more hairy,
Cecil Westerhof wrote:
That will only work when every user is in a group.
I think that there is a way to get the group(s) of an user. When the
only requirement is that the person is part of a group, you could
check that it is not empty.
Maybe you mean identity.conditions.in_any_group()
Lukasz Szybalski wrote:
On Tue, Jun 10, 2008 at 10:05 AM, Cecil Westerhof
[EMAIL PROTECTED] wrote:
2008/6/10 Lukasz Szybalski [EMAIL PROTECTED]:
I think that there is a way to get the group(s) of an user. When the
only requirement is that the person is part of a group, you could
[EMAIL PROTECTED] wrote:
make TG happy because it needs an ORM for the identity and permission
tables.
So I can just SQLAlchemy for the Identity and Permissions stuff and my
own for everything else?
Yes, and I recommend you looking at the two lower layers of SQLAlchemy,
instead
Lukasz Szybalski wrote:
class th(object):
pass
How do I do the same query in turbogears?
th.query.filter(...).all()
As far as I know the session is managed by turbogears? What exactly happens
with the session?
The session is bound to the mapped classes.
It's just one way to
Vortexmind wrote:
Instead of patching it to add a check for the existence of __json__, I
use to comment out support for sqlalchemy in turbojson.
I don't want every single column that is stored in the DB to be
serialized down to the user's browser.
There is, like, security, privacy issues
Vortexmind wrote:
It's a tradeoff between ease of use and complexity, not so simple.
While I'm perfectly fine with the way it is now (I just have to monkey
patch turbojson), I think a mixin class (a la zope) or a register_json
function could have been equally simple to use.
For
Vortexmind wrote:
If I'm right, this is what I'm talking about ...
http://trac.turbogears.org/ticket/1619
Instead of patching it to add a check for the existence of __json__, I
use to comment out support for sqlalchemy in turbojson.
I don't want every single column that is stored in the
Christoph Zwerschke wrote:
So the tg_errors argument has been passed correctly and the dummy
variable still contains the Root object. This made me believe I need
the dummy variable.
I don't understand fully what's going on especially because Chris
suggested that the dummy variable is not
But I thought that the use of innerHTML was frowned upon, or is that
not the case?
There are different opinions.
http://www.robertnyman.com/2006/04/20/we-all-love-innerhtml/
The argument is simple enough:
- Calling a URL (with proper template and stuff) and inserting that output
Christoph Zwerschke wrote:
validator = validators.NotEmpty)
I think you must instantiate validators, i.e. use validators.NotEmpty().
From formencode.declarative:
Declarative objects for FormEncode.
Declarative objects have a simple protocol: you can use classes in
lieu of
Diez B. Roggisch wrote:
This is the expected behaviour - it is the
formencode-parameter-conversion that works that way. And AFAIK that was
the same in TG0.9
If you want the original paramters, inspect the cherrpy.request.params.
cherrypy.request.params is already nested, at least
Marco Mariani wrote:
cherrypy.request.params is already nested, at least here.
You might want to call variable_decode from FormEncode.
Ehr, variable_encode, that is. That's what I do.
--
This e-mail (and any attachment(s)) is strictly confidential and for use only
by intended
Vortexmind wrote:
I'm new to Turbogears: I've chosen it because of the supports it
provides (with SQLAlchemy) for tables with multiple field primary
keys. I have a legacy database to use, and I need to know if there is
sort of automatism to import model from legacy database schema, or
Diez B. Roggisch wrote:
http://docs.turbogears.org/1.0/UnifiedControllers
It's quite useful but unfortunately buried in the documentation.
OMFG. That's one piece of ugly code.
Wait, wait!
I am trying to get rid of a similar Piece Of Ugly Code.
I feel guilty for having used that,
shday wrote:
Another tid bit. During development, reflection can be a PITA because
it can increase application startup time considerably. This can be
overcome, however, by pickling your reflected metadata and using that
(instead of reflecting every time).
My current application reflects
Diez B. Roggisch wrote:
But also (and as you say yourself) it does not rid you of knowing JS.
I think the most common reason given (by programmers) for not using
javascript is its perceived quirkness and the scarcity of features in
the standard library.
As one bright Python developer
John M Camara wrote:
One thing no one has mentioned so far is that there will be no support
for a relational database. I personally think this is a good thing so
people can realize that other datastore options are available.
It would have been better if the query language didn't have a
Diez B. Roggisch wrote:
In the end, programming is and will always be that - programming. So I'm
really having difficulties to imagine that more than rather simple usecases
can be covered.
Instead, I prefer a lib like MochiKit or jQuery that makes the unavoidable as
nice and smooth.
Matt Wilson wrote:
Glauco, thanks for the feedback. What keyword do I use to pass this
into my TableForm? This doesn't look like a complete template, so I
don't think I should pass this in with a template keyword.
The code Glauco posted is from the TableForm's template. Look for
Christopher Arndt wrote:
http://chrisarndt.de/projects/fancyflashexample/
oops.
The requested URL
/projects/fancyflashexample/FancyFlashExample-1.0-py25.egg was not found
on this server.
--
This e-mail (and any attachment(s)) is strictly confidential and for use only
by intended
Daniel Fetchinson wrote:
I haven't looked at your code yet but have to say that I also don't
like very much the fact that I have to define what widgets are used in
the controller. It feels more natural to declare this in the template.
I don't claim to have a very precise opinion on what
Richard Jones wrote:
I'm at a loss how to configure TG to actually have a URL of
submit.php.
If the problem is the dot, declare the controller as submit_php, it
should work AFAIR
--
This e-mail (and any attachment(s)) is strictly confidential and for use only
by intended recipient(s).
Diez B. Roggisch wrote:
I was expecting the default mochikit file from the TG distribution not
be included in the resulting web pages, but it is. I don't see any
changes related to the suppress flag. Where should I be able to see
the change if the flag is on or off?
That's strange.
Florent Aide wrote:
- Second the communication on where the 1.0 and 1.1 branches are
going as been quite feeble (to say the least) and this is my own
fault. I'll need to amend my ways. I promise I'll try to find time on
new year's eve to communicate more on the planning, the milestones and
Frank Slotta wrote:
I hope this could be useful, please send suggestions if something is
wrong or obsolete.
Seems fine, although with the current release you can also point
identity.provider to a class path, and it will be loaded with load_class()..
This way, there is no need to
Kevin Cole wrote:
I don't care how many columns are returned, but I want to tell it
which ones should be distinct, and still get the benefits of the
mapping.
What do you mean which ones should be distinct ?
I mean, distinct is not a function (*)
What's your use case?
(*)
Javier Rojas wrote:
I apologize right now if there are any grammar errors or things like
that; my mother language isn't english :)
There is at least one link error
YUI my opionion links to javascript-libraries#my-opinion
and it says I hate not having proper documentation.. in
vimal wrote:
?python
if usr_rght in user_rights: //usr_rght is an
integer and user_rights is a list//
add_user = false
else:
add_user = true
?
If you can avoid ? snippers, please do yourself a favour.
input type=button value=ADD USER
vimal wrote:
script type=text/javascript
function onClick()
{
window.location.href=(/student,{'name':${name},'age':${age}}
I guess you need to compose the full URL - and include ${name} in quotes.
function queryUri(base, params) {
for (var key in params) {
if
Jason ha scritto:
greetings all I know this is going against the MVC structure
Come on. If it makes sense to do that, you should :-)
but I need to send a sting variable from the controller to kid for display
it without it all getting converted in special HTML characters.
thanks!
You
[EMAIL PROTECTED] ha scritto:
print kid.Template('span xmlns:py=http://purl.org/kid/ns#;
value=${None} /')
?xml version=1.0 encoding=utf-8?
span /
But just looking it at I would assume it would suffer from the same
problem.
I am not sure I understand your concern.
No readonly
Kevin Cole ha scritto:
input type=text readonly=$readonlyvar/
Wouldn't it be better to make $readonlyvar equal the entire string
readonly=readonly
That would defeat the whole point of having an XML based template engine.
I'm used to that since TAL was introduced in Zope 2,
Halldór ha scritto:
Hi there,
How do I send headers with Turbogears (or it's probly Cherrypy)...
cherrypy.response.headers['Expires'] = '...'
cherrypy.response.headers['Cache-Control'] = 'no-store, no-cache,
must-revalidate, post-check=0, pre-check=0'
cherrypy.response.headers['Pragma']
Thierry ha scritto:
Is there a way to turn off the strict xhtml parsing?
From app.cfg:
# which view (template engine) to use if one is not specified in the
# template name
# tg.defaultview = kid
# The following kid settings determine the settings used by the kid
serializer.
# One of
exhuma.twn ha scritto:
static/icons/theme1/*.png
static/icons/theme2/*.png
static/icons/theme3/*.png
You can register an alieased path in app.cfg:
[/themed_icons]
static_filter.on = True
static_filter.file = %(top_level_dir)s/static/icons/theme1
then you refer to them as
The Alchemist ha scritto:
I am working with TurboGears 1.0.3.2 and have downloaded TGYUI 0.1.1
I need help to create a widget (say, a grid) using TGYUI, and then
seeing it in the browser (through an exposed method in the
controller).
Is there documentation for using the TGYUI package? Or
krsyoung ha scritto:
Is there some way I can make a common_fields subclass that I can then
reuse in both of the approver_fields and admin_fields? Basically I
want to pull out name so I don't have to repeat it for each widget.
WidgetLists are lists, you do not subclass them; you add
jype ha scritto:
I _think_ this is MochiKit 1.3.1's fault - the form-value-gathering-thingy
isn't working properly.
Try switching to MK 1.4, or rip 1.4's form gathering code (don't recall the
name right now) and patch that into the RemoteForm.
I tried that but it didn't work. I
Florent Aide ha scritto:
I am trying to make a small TG application, which should save PDF file.
It stores letters in a PDF format in a relational database.
Do you believe it is a good idea to use a class as below? Any idea welcome.
class Letters(Entity):
Diez B. Roggisch ha scritto:
I'm not aware that anti-aliasing is available. There is no way for drawing
2D,
let alone 3D unless you use flash or SVG.
I'm sure there is a way, and it seems anti-aliased to me:
http://www.abrahamjoffe.com.au/ben/canvascape/
Lee Connell ha scritto:
That worked but when I put my identity model in its own it broke the
identity.
If you move the identity classes, you must tell TG the new location in
app.cfg or dev.cfg:
visit.saprovider.model = myproject.model.Visit
identity.saprovider.model.user =
Lee Connell ha scritto:
What do you mean, what didn't work for me?
Did you try splitting the model? Did it work?
Why do I need to move model.py into __init__.py
To avoid changing the code in controllers all at once.
couldn't I move whatever
I had in there into model/new_model.py?
iain duncan ha scritto:
Thanks, that's good info. What do you guys use to profile on a page by
page basis?
Aside from python profiling... on the client side, you can try Google's
Load Time Analyzer
https://addons.mozilla.org/it/firefox/addon/3371
Lukasz Szybalski ha scritto:
There is around 20 tables with a lot of foreign key, compound keys and
alternative keys.
Which mapper do I use?
By mapper, do you mean a layer like ActiveMapper or Elixir?
I am happy with the autoload feature. No need to declare the tables at all.
Actually,
Lee Connell ha scritto:
Is there anyway to split my model up?
Sure. Move your model.py into model/__init__.py, and add more modules.
What didn't work for you?
I would like to split each table/entity into it's own class
its own module, you mean.
Glauco ha scritto:
Whats the best solution for validate a RepeatingFieldSet?
from turbogears.validators import Schema, ForEach, String, Int
from turbogears.widgets import WidgetsList, RepeatingFieldSet, TextField
Say you have a repetition of
class MyFormFields(WidgetsList):
stuff =
Chris Curvey ha scritto:
I can't believe this is not a FAQ,
Yes, it is.. googling sql tree gives relevant results on both web and
newsgroups.
If I have a hierarchy of categories, and my model looks like this:
class Category(SQLObject):
name = UnicodeCol(length=100)
parent =
nossatv_sd ha scritto:
I have seen people using sha package to encrypt strings (e,g
passwords). An example is shapassword =
sha.new(kw.get('password')).hexdigest(). Now I am not sure how to
decrypt the password encrypted this way
This is not encryption.
Please, please have an idea of what a
Richard Clark ha scritto:
Uncalled for.
Probably, or maybe I should've explained myself a little more.
Oh, come on, there was a double smiley :-)
The vast majority of web developers have no idea what a
hash function is, and have no need to know. That's why identity Just
Works. The day
johnbraduk ha scritto:
People never seem to learn that if it ain't broke don't fix it.
John
With all due respect to Ian Bicking, SQLObject 2.0 is broken in some
way, or we would already have a 2.0 release:
http://blog.ianbicking.org/sqlobject-2.html
This is without going deep in the
saliez ha scritto:
[...]
* However I do not understand the UBUNTU philosophy about the
management of external packages:
I understand well they cannot provide full support for all external
packages, particularly when they are continuously under developements.
The problem of the current
Gaetan de Menten ha scritto:
What's worse (from an upgrade point of view) is that many query
methods are also deprecated (though still present in 0.4), namely
get_by, select, select_by, etc... Most of these should be replaced by
a filter_by() call. For example, MyClass.get_by(name=test)
I've just switched my ftests from twill+mechanize+... (too complex for
my taste) to testutil.BrowsinSession.
It seems BrowsingSession does not follow redirects, like the previous
setup did.
So, I have to manually call the new URL from the test methods..
I tried naively fixing the issue in my
Jorge Godoy ha scritto:
Is there any of those products that allows printing the charts on a standard
install of Firefox / IE? 'cause I had a client who didn't buy one of those
charting libs because he couldn't print them (and telling all his clients to
change something was not
At first, it was strange that all my fancyflash() messages stopped
working...
then I traced it back to the session_id cookie, that is set (or cleared)
but not kept on Firefox between requests.
So, this controller:
@expose()
def session_id(self):
import cherrypy
Shan ha scritto:
SessionNotEnabledError: bound method Root.index of
shan.controllers.Root object at 0x917620c
do you have
session_filter.on = True
in your config?
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google
First thing, I'd like to thank you for the release.
Second thing, I'd also love if somebody could help me understanding why
all my unittests are failing after the upgrade :-))
Basically, it seems like the transactions are not rolled back.
I use SQLAlchemy. I create the test db, complete
iain duncan ha scritto:
I would like to add the option for my client to download ready to print
invoices as one big doc, and am wondering what people would use to do
that. I have heard good things about the Apache POI project, but it
seems to have only Ruby bindings. Anyone have any
I'm currently doing some more integration of YUI widgets in TGWidgets,
on top of Mr Kubacki's package,
and I wonder if there are best practices (or framework mechanisms I've
overlooked) to..
1) have an 'id' property much like the field_id in FormField, to be
able to properly refer to the
Jorge Godoy ha scritto:
For Excel you can use CSV and it will open it.
Sure.
Except, it will try and cast the strings to number, dates, and the like,
depending on content... you might not like it.
Openoffice or KOffice are no better, and will try and cast the types in
a different way.
If
Jorge Godoy ha scritto:
1) have an 'id' property much like the field_id in FormField, to be
able to properly refer to the widget container in YUI's constructors and
CSS
Hmmm... You can override the mechanism used on widgets to do that. But you
have to be careful because it is
On Thu, 12 Jul 2007 18:21:40 +0200, Diez B. Roggisch
[EMAIL PROTECTED] wrote:
def get_link(o):
el = Element('a',href=tg.url('/path/to/view/',object=o.id)
el.text = o.description
return el
This should be easier doable using the KID XML-function:
def get_link(o):
return
Alberto Valverde ha scritto:
There are also many others which I haven't mentioned but have not
helped less and I'd like to publicly thank them now (we know who you
are! the changelog never lies ;)
This is a welcome status update, thank you.
BTW, will some of the TG developers be in
Sanjay ha scritto:
thing.save()
session.flush()
I use just session.flush() or thing.flush(). What is thing.save() used
for?
It's needed when your mappers are not implicitly attached to a session.
If you use assign_mapper like most turbogearers, save() is useless.
Jesse James ha scritto:
Which handler are you using?
http://docs.python.org/lib/node410.html
RotatingFileHandler.
...which inherits from FileHandler, which inherits from StreamHandler,
which is buffered.
The open() is in FileHandler.__init__... there is 'mode' and 'encoding'
Florent Aide ha scritto:
This does not mean that I don't use other libs for display effects...
ever looked at qooxdoo? (http://qooxdoo.org/)
Did you use it? Is the API clean enough?
From my 20 seconds inspection, it seems lighter than dojo.
Does it play well with the mighty pythonic
willhardy ha scritto:
PS: if you are a member of digg or Slashdot and would like to support
the survey, here are the relevant articles:
http://digg.com/programming/How_does_your_web_development_platform_rate
http://slashdot.org/firehose.pl?op=viewid=205681
is this a joke or what?
Marco Mariani ha scritto:
PS: if you are a member of digg or Slashdot and would like to support
the survey, here are the relevant articles:
http://digg.com/programming/How_does_your_web_development_platform_rate
http://slashdot.org/firehose.pl?op=viewid=205681
Sylvain Hellegouarch ha scritto:
I don't _NEED_ two different platforms, do I?
Maybe the question is not correctly asked but this happens quite often
that you have to work on different environment. Well it happens to me
anyway.
Different python environments, yes. And some amount
iain duncan ha scritto:
Thanks Alaa. I will try sqlalchemy and genshi. I'm not far enough
down a path with either sqlobject or kid to feel indebted to using
those going forward.
I would second that suggestion. If you start down the road with TG using
Genshi, SA, and maybe even
Sanjay ha scritto:
As I understand from the commented documentation in app.cfg (see
below), if I assign a text password to any user object like this:
u.password = 'clear text'
No.. as you can see in test_identity.py, the password content is meant
to be already encrypted.
def
Daniel Fetchinson ha scritto:
nice shot! I've put this into TG FAQ: http://docs.turbogears.org/FAQ#q-033
Well, care is needed if the name of the project also appears as
variable/class/method/etc names which might very well be the case.
This recipe could make a mess for somebody
[EMAIL PROTECTED] ha scritto:
That's an interesting suggestion. I'm not a CherryPy expert, so how
do I bind something to a request like that?
This way:
def get_flash_service():
try:
ret = cherrypy.request.flash_service
except AttributeError:
ret =
[EMAIL PROTECTED] ha scritto:
Hi everyone,
I've done some more work on my alternative to turbogears.flash(), and
it's pretty decent as this point (at least I think so), and it might
be useful for others.
Thank you, it's just like what I needed, the way I would have done it :-)
I
1 - 100 of 180 matches
Mail list logo
