I have several custom page content types of the form:

class MyWidget(models.Model):

    ...fields...

    class Meta:
        abstract = True
        verbose_name = _('my widget')
        verbose_name_plural = _('my widgets')

Page.create_content_type(MyWidget)

If I add these to a page, and later try and delete the page I get the IntegrityError exception:

IntegrityError at /admin/page/page/82/delete/ (1451, 'Cannot delete or update a parent row: a foreign key constraint fails (mydatabase.page_page_mywidget, CONSTRAINT parent_id_refs_id_54aa763d FOREIGN KEY (parent_id) REFERENCES page_page (id))')

I'm assuming this is due to an inappropriate Django ondelete handler on FK field for the page_page_mywidget table autogenerated by Page.create_content_type().

Ideally, when a page is deleted, any rows in content type tables linked to the deleted page should also be deleted. I believe this corresponds to the CASCADE ondelete behavior.

A slow and frustrating workaround is to manually delete all content types on the page, save the page, and then delete the page.

I'm having a constant problem. I've updated to the latest version (ab635ef2059b57b260547dc5bf8bdb3ade3247d3) and it's still not working. The site is running with PostgreSQL. When I edit text in a richtext item in my page, and I clicked save, it's not updating. The text just goes back to what it was before. This has been a problem on my WebFaction site, running PostgreSQL for a good few weeks and even updating to latest version still isn't fixing it!

I am trying to register a content type on two models:

ModelOne.register_regions(('main', 'Main'))
ModelOne.create_content_type(MyContent)
ModelTwo.register_regions(('main', 'Main'))
ModelTwo.create_content_type(MyContent)

But I am receiving an error from the final line:

ImproperlyConfigured: Cannot create content type from non-abstract model (yet).

This is because MyContent is getting redefined in my models.py on the second line.

I must admit I am finding this one very hard to debug, but I suspect it is caused by one of the calls to type() in feincms.models.

The attached diff has a failing test for the problem.

Do you know what may be causing this? I would love to help if I can, but I am a little stumped as to what action to take.

I'm having a lot of trouble getting the blog module working. If i register a content type on Entry, ie.

Entry.register_regions(
    ('main', 'Main region'),
)
Entry.create_content_type(RichTextContent)
Entry.create_content_type(MediaFileContent, POSITION_CHOICES=(
    ('left', _(u'left')),
    ('right', _(u'right')),
    ('center', _(u'center')),
    ))
Entry.create_content_type(ImageContent, POSITION_CHOICES=(
    ('left', 'Default position'),
    ('right', 'Default position'),
    ('default', 'Default position'),
    ))

This throws me a lot of errors when starting the dev server.

RuntimeWarning: Cannot create content type using feincms.content.richtext.models.RichTextContent for feincms.module.blog.models.Entry, because richtextcontent_set is already taken.
RuntimeWarning: Cannot create content type using feincms.content.medialibrary.models.MediaFileContent for feincms.module.blog.models.Entry, because mediafilecontent_set is already taken.
RuntimeWarning: Cannot create content type using feincms.content.image.models.ImageContent for feincms.module.blog.models.Entry, because imagecontent_set is already taken.

How can i remedy this situation? I keep the page module stuff in an app called pages and the blog related stuff in app called blog.

Any help will be greatly appreciated!

We've just started setting up a FeinCMS based blog, but have encountered a small problem.

We set up the Page model in the normal way, and register a few extensions like this:

Page.register_extensions(
    'feincms.module.extensions.datepublisher',
    'feincms.module.page.extensions.excerpt',
    'feincms.module.page.extensions.relatedpages',
)

The extensions appear to only be loaded once - that line of code is run once, no "AlreadyRegistered" exceptions, and the migrations all work fine which implies the model is being created properly.

However, the admin shows two tabs for each extension. The tabs contain the same content, but only one (non-deterministic) tab will work at any time, which means you have to enter data twice in the admin to be sure that it will save. Also some widgets in the repeated tabs just don't work, an example of this is the multiple selection widget for the "relatedpages" extension.

I don't know enough about FeinCMS to know where to even begin looking for what might be causing this. Does anyone have any ideas about what might be causing this, or even a good place to start looking?

If I have a FeinCMS Base subclass MyDocument, and in the same module a FeinCMS content_type (or indeed any class) called MyDocumentContent, this results in database weirdness, because FeinCMS generates a class with the pattern "%sContent".

Assuming the "%sContent" can't be renamed to something less clashable, then I suggest clashes should be caught and warned against.

Using latest version from github. Django 1.2.1, with a sqlite3 database. It doesn't seem to happen all the time, but a lot of the time, the stuff I'm writing into the rich text editor isn't saving. It says it's saved, but the text reappears just as it was before. I have to delete the rich text editor from the region and then re-add it to then get it to save. ?

Feincms (version 1.18.0) build-in page modules (probably not only them) dont work with Django 3.1.

Duaring the doc (https://feincms-django-cms.readthedocs.io/en/latest/page.html#) "page" field in admin panel returns empty page.

Temporary solution: uninstall django 3.1 and install django 3.0

We encountered a couple of problems when using the medialibrary app with a file system other than the default django.core.files.storage.FileSystemStorage.

When using storages.backends.s3.S3Storage from django-storages MediaFileBase raised an error because it was trying to initialise S3Storage with the kwargs "location" and "base_url". S3Storage is a remote storage using Amazons S3 storage so has no concept of base_url.

We fixed the issue by catching the exception and initialising the storage with no arguments. In addition the S3Storage does not implement the path property.

It turns out that threading.local on Apache wsgi is persistent over multiple requests which it should not. This turns out to be an issue with the app_reverse function not working properly for web sites which have the same application content in several pages. Usually multi language sites.

The issue has come up in this ticket: https://github.com/feincms/feincms/issues/306 but that one was about something different. So let's keep the discussion about this issue in this ticket here.

I'm suggesting getting rid of threading.local and instead use the cache backend for app_reverse caching.

Just discovered that including 'feincms.urls' at a non-blank point the URLconf (e.g. '/content/') stopped working in an upgrade from 1.4.2 to 1.6.2.

url(r'^content/', include('feincms.views.legacy.urls')), works correctly

Losing this functionality seems like a backwards development?

Also discussed here: https://groups.google.com/forum/?fromgroups#!topic/django-feincms/vFwfGkYdbQg but the suggested solution doesn't work (even after adding .as_view(), the handler doesn't seem to cope with not being at the URLconf root)

Django 3 docs say (and Django 4 actually removes the feature):

To limit creation of sessions and hence favor some caching strategies, django.views.i18n.set_language() will stop setting the user’s language in the session in Django 4.0. Since Django 2.1, the language is always stored in the LANGUAGE_COOKIE_NAME cookie.

We should adapt feincms/extensions/translation.py as well as the tests that rely on that behaviour?

It's not super-urgent, as the fallback for django1.6 also somewhat covers the dj4 case, but that's more a happy accident than intentional.

Context: When creating a FeinCMS site using some kind of plugin mechanism, I ran into this problem.

Each plugin wants to contribute some Template to the main project. To do that, it will do something like

from feincms.module.page.models import Page

regions_main = (('main', 'Content'), )
Page.register_templates(
    {'key': 'mmmmh', 'title': 'Tasty template', 'path': 'munch/nomnom.html', 'regions': regions_main}
)

This will not work.

Each region holds a list of content types that can be used in that region. Unfortunately, this list is built when a concrete content type is created, so calling register_templates again later will add regions that have no content types at all.

I can paper over this by patching in all content types here

for region in Page._feincms_all_regions:
    if not region._content_types:
        region._content_types = Page._feincms_content_types.copy()  # copy is important otherwise CTs will be duplicated

but that is not always what is wanted, since I can't restrict CTs to a region that way.

If we ever want to restrict a CT to a certain region, things will become more complicated (the region set for each CT is not stored anywhere, so that information is lost once the CT has been instanciated).

Food for thought...

Hi folks,

This is between an issue and a feature request I guess. Here is the use case:

• You set up a page to an url, delivering a customer request.
• The customer has an internal discussion and the content needs to be moved/renamed
• You rename/move the content to some other location
• The old content is still indexed by some pages or search engines, but lead to a 404

Ideally, there would be a module/plugin that would automatically add 301 redirects to the old urls. Optionally, there is some option to do so while renaming/moving.

What do you think about it @matthiask ? Is there a work-around, or some way to do so currently? I couldn't find anything like this in the docs, neither browsing the code

I would like to override a few methods in PageManager - what would be the most efficient way to achieve that?

Here's some context to explain why I want to do that.

I wanted to use Django's Sites framework, and configure FeinCMS in a way that would allow me to control on which sites any particular page can be shown.

Yes, FeinCMS already has an extension for that: https://github.com/feincms/feincms/blob/e923d0466b74937d36eb45e2c46ea18cfef83d56/feincms/module/page/extensions/sites.py

However:

1. It creates 'site' ForeignKey (I want to have 'sites' ManyToMany instead, but I easily fixed that by writing a custom extension)
2. It relies on Site.objects.get_current()

Site.objects.get_current() works only if you have SITE_ID defined in your settings, otherwise it raises an error and asks to provide request in order to deduce the current Site. I omitted SITE_ID from my settings on purpose because I want to achieve behavior documented in Django docs: https://docs.djangoproject.com/en/dev/ref/contrib/sites/

If the setting is omitted, the get_current_site() function will try to get the current site by comparing the domain with the host name from the request.get_host() method.

There are a few problems, though. I noticed that feincms sites extension modifies which pages are treated as active by using the following approach:

def current_site(queryset):
    return queryset.filter(site=Site.objects.get_current())

PageManager.add_to_active_filters(current_site, key="current_site")

However, without SITE_ID specified in settings, the only way to filter by current site is to somehow pass request to the filtering logic, and from my understanding, with add_to_active_filter there is no way to do so.

Which brought me to conclusion that I need to research how Handler view works, because views have access to request.

I've tried many different approaches, but the only one that worked was to copy FeinCMS Handler code and then manually redefine PageManager methods before Handler class is defined. https://github.com/feincms/feincms/blob/e923d0466b74937d36eb45e2c46ea18cfef83d56/feincms/views/init.py#L29

from feincms.module.page.models import PageManager
from .managers import for_request, page_for_path, best_match_for_path

PageManager.for_request = for_request
PageManager.page_for_path = page_for_path
PageManager.best_match_for_path = best_match_for_path

logger = logging.getLogger(__name__)

class Handler(ContentView):
    # ...

custom for_request simply passes request to custom page_for_path and best_match_for_path

page_for_path and best_match_for_path are overridden in order to accept request as an argument and to additionally filter pages with .filter(sites__in=[request.site])

Although it's very cumbersome, it actually works.

I tried other, seemingly more elegant approaches, but they didn't work:

1. If I create a custom manager and try to redefine objects in Page, then AttributeError: 'NoneType' object has no attribute '_mptt_meta' is thrown in django-mptt. It's probably because of all the MetaClass magic. I assume there must be some way to fix that?
2. I can override FEINCMS_DEFAULT_PAGE_MODEL, but it creates many unwanted side-effects (i.e. I will have to manually recreate Page admin to import custom Page model, because current admin.py in FeinCMS does not use FENCMS_DEFAULT_PAGE_MODEL). Even when I did that, I had to rename db_table in the custom Page.Meta class, after which django was still complaining about two different models named Page. I ended up ditching this approach because too many things started falling apart.
3. If I use add_to_class('objects', PageSitesManager()) in my custom extension, where PageSitesManager is the custom Manager will all the important methods overridden, it simply does nothing.

Is there a more elegant/obvious way to override PageManager without causing errors in django-mptt?

If you add a new page and change the template, you can click on one of the save buttons (_addanother, _continue, _save) before the page if fully loaded, which results in 2 pages being created.

This behavior could be prevented, if you disable all the save buttons right after the line: form_element.find('[type=submit][name=_save]').click(); in the on_template_key_changed function in item_editor.js or you could render a loading screen to prevent the user from working on the page.

On my install, I find basic article pages take about 5500 queries to produce, and take around 200 seconds to load.

Specifically, this query is repeated about 2000 times:

SELECT "medialibrary_mediafiletranslation"."id", "medialibrary_mediafiletranslation"."parent_id", "medialibrary_mediafiletranslation"."language_code", "medialibrary_mediafiletranslation"."caption", "medialibrary_mediafiletranslation"."description" FROM "medialibrary_mediafiletranslation" WHERE ("medialibrary_mediafiletranslation"."parent_id" = 802 AND (UPPER("medialibrary_mediafiletranslation"."language_code"::text) LIKE UPPER('en-us%') OR UPPER("medialibrary_mediafiletranslation"."language_code"::text) LIKE UPPER('en%'))) ORDER BY "medialibrary_mediafiletranslation"."language_code" DESC LIMIT 1

This one around 2000 times as well:

SELECT "medialibrary_mediafiletranslation"."id", "medialibrary_mediafiletranslation"."parent_id", "medialibrary_mediafiletranslation"."language_code", "medialibrary_mediafiletranslation"."caption", "medialibrary_mediafiletranslation"."description" FROM "medialibrary_mediafiletranslation" WHERE ("medialibrary_mediafiletranslation"."parent_id" = 965 AND (UPPER("medialibrary_mediafiletranslation"."language_code"::text) LIKE UPPER('en-us%') OR UPPER("medialibrary_mediafiletranslation"."language_code"::text) LIKE UPPER('en%'))) ORDER BY "medialibrary_mediafiletranslation"."language_code" DESC LIMIT 1

And this one around 1500 times:

SELECT "medialibrary_mediafiletranslation"."id", "medialibrary_mediafiletranslation"."parent_id", "medialibrary_mediafiletranslation"."language_code", "medialibrary_mediafiletranslation"."caption", "medialibrary_mediafiletranslation"."description" FROM "medialibrary_mediafiletranslation" WHERE "medialibrary_mediafiletranslation"."parent_id" = 346 LIMIT 1