On Thursday, January 22, 2015 at 10:59:28 AM UTC+1, Gael Princivalle wrote:
>
> Thanks Niphlod I understand that the web2py grid cannot do all that users
> would like to, but for my use it's quite to.
> Viewing, sorting, editing, creating new rows, I like it.
> For my needs it just miss a way for sorting rows properly by fields that
> contain an row id of another table.
> If there's a way to sort and search by the represent function or format
> instead of id I think that we will have more the possibility to use the
> web2py grid.
> I'm afraid that I don't understand your explanation about how to do it.
> The represent function don't change anything, almost as I've done it.
> If you have a suggestion thank you.
>
>
simply put, web2py can't ATM create a single query that includes by default
all the fields needed for your representation and make it work in the db.
Take this ( corner case )
db.define_table('a_table', Field('a_field', represent = lambda value, row :
"this
is a test message %s" % row ))
or, even better
import uuid
db.define_table('a_table', Field('a_field', represent = lambda value, row :
uuid.uuid4() ))
nice... perfectly valid representation.... but how can you build a query to
search on the backend for something that in the database doesn't exist
(with the added value that it changes every time you see it) ?
referenced records are a subset of the representation issue, but hold the
exact same issue: web2py can't build a query on the table because it just
knows how to render a SINGLE row, not the whole set. To make grid search
work, the grid would need to:
- make a query on the first table ("master")
- make a query for every single row in the "master" table
- keep all of those in memory
- filter/sort in python
- paginate
it'll make your website crash (or at least render it sluggish as hell) in
two seconds straight.
Fortunately, in some cases, you will be able to construct a grid with a
model that involves SIMPLY reference representation that works fine....
Given your "original" simple model
db.define_table('l1_categories',
Field('name'),
format = '%(name)s')
db.define_table('l2_categories',
Field('l1_category', 'reference l1_categories'),
Field('name'))
this will let you search/order/etc correcly
q = db.l2_categories.l1_category == db.l1_categories.id
fields = [db.l2_categories.id, db.l2_categories.l1_category, db.
l1_categories.name]
grid = SQLFORM.grid(q, user_signature=False, fields=fields, field_id=db.
l2_categories.id)
and edit the l2_categories with the "usual" dropdown on the l1_category
field.
The secret is that you're swapping the record representation with the
actual value stored in the database, and that the representation matches
exactly the content of the field "pulled" from the other table. passing a
field_id to the "original" table lets you edit the records pertaining to
that table, as usual.
--
Resources:
- http://web2py.com
- http://web2py.com/book (Documentation)
- http://github.com/web2py/web2py (Source code)
- https://code.google.com/p/web2py/issues/list (Report Issues)
---
You received this message because you are subscribed to the Google Groups
"web2py-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
