I just implemented a GZipMiddleware for Django, so now you can enjoy the bandwidth-saving goodness that is gzip without using django.middleware.cache (actually, since Djangos cache-middleware is oblivious to the Vary header, it will not work with GZipMiddleware…)

It is in ibofobi.middleware.gzip under http://mel.ibofobi.dk/~sune/r/ibofobi.git. Enjoy.

Yay!

I finally got around to implementing a simple commenting system here.

Please, pretty please, don’t let the spammers find me…

The XHTMLAsHTMLMiddleware class now sets (or updates) the Vary response-header, to indicate to HTTP-caches that the response might change, based on the content of the Accept request-header.

Now, I just need to hack the Django cache-middleware, to set, understand and obey the Vary-header, so I can use it with XHTMLAsHTMLMiddleware.

I’ve written two gadgets for helping me when developing Django sites. Since they might be useful to others, I thought I would mention them here.

ibofobi.utils. developer_http

The first is a web-server which re-exec’s itself after every request, so that all resources are reloaded for every request. It has worked very well for me. It lives as ibofobi.utils.developer_http.py in http://mel.ibofobi.dk/~sune/r/ibofobi.git.

You can either execute it directly, with --bind and --port (it defaults to all interfaces port 8080). Or if you just want to WSGI-server in there you can import BaseWSGIServer from the module.

ibofobi.middleware. template_dirs_hacker

The second gadget is a middleware which munges django.conf.settings.TEMPLATE_DIRS for every request. This is how I use it in my settings.devel:

from main import *
import admin

ROOT_URLCONF = 'settings.urls.devel'
MIDDLEWARE_CLASSES = ('ibofobi.middleware.template_dirs_hacker.TemplateDirsHacker',) + MIDDLEWARE_CLASSES
ADMIN_MEDIA_PREFIX = admin.ADMIN_MEDIA_PREFIX
TEMPLATE_DIRS_MAPPING = (
    ('/admin/', admin.TEMPLATE_DIRS),
    ('/', TEMPLATE_DIRS),
)
TEMPLATE_DIRS = []

And in my settings.urls.devel I have:

(r'^admin/', include('django.conf.urls.admin')),
(r'', include('settings.urls.main')),

So, when I request /admin/ I get the admin site with the proper TEMPLATE_DIRS, and when I request anything else I get the main site with the TEMPLATE_DIRS for that.

I like to serve my pages as XHTML, and with the correct content-type header application/xhtml+xml when possible, and downgrade to text/html for XHTML-challenged browsers. Out of the box Django cannot do this, but luckily the middleware architecture comes to my rescue :)

So, since this is such a short and sweet piece of code, here it is:

import re

re_ct_xhtml = re.compile(r'^application/xhtml\+xml\b')
re_accept_xhtml = re.compile(r'\bapplication/xhtml\+xml\b')

class XHTMLAsHTMLMiddleware:
    """I change content-type application/xhtml+xml into text/html, if the
    browser does not support the XHTML content-type."""

    def process_response(self, request, response):
        if re_ct_xhtml.match(response['Content-Type']):
            accept = request.META.get('HTTP_ACCEPT', '')
            if not re_accept_xhtml.search(accept):
                ct = response['Content-Type']
                ct = ct.replace('application/xhtml+xml', 'text/html')
                response['Content-Type'] = ct

        return response

It is also available as ibofobi.middleware.xhtml in http://mel.ibofobi.dk/~sune/r/ibofobi.git.

I’ve just converted this site to Django. In the process I wrote a few small apps, two of which are in a state, where others might use them (or improve them, until they become useful :).

The first app is a blog, with posts and categories (which drives this blog…). It’s still not feature-complete (I want comments and drafts at the very least), but it works. One nice feature, is that the posts are formatted using Markdown.

The second app is a personal calendar-type-thingie, it’s stille very rough around the edges, but it helps me to keep track things I have to remember…

They live in http://mel.ibofobi.dk/~sune/r/ibofobi.git under src/apps/, I still haven’t gotten around to writing a setup.py, and the templates need a bit of cleaning up.

I use rss2email to read blogs of various people. Combined with gmail rss2email is a brilliant way to read blogs. But, for a long time I have missed an easier way than ssh ibofobi.dk r2e add http://some.blog when adding new feeds. So, I decided to write a quick little rss2email web admin-interface. You can get it at http://mel.ibofobi.dk/~sune/r/r2eadmin.git.

As a bonus, here is a bookmarklet to subscribe; you just have to edit the URL.