Python Desktop Server Weblog 2003/5

Patches from Garth (MeshTool, WikiTool) are in CVS

Garth latest patches are now included in CVS. So don't be shocked if you update from CVS, as there are lot's of changes. Try out his WikiTool, it's cool!

Garth is working on better stray link resolving

Garth has a new idea to factor out the stray link resolving that came with the CVS 0.5.0 version to it's own tool to have even more fun with it. Sounds good, jump over to read the proposal and comment on it.

I think I will wait for this refactoring before putting out a 0.5.0 version.

Version 0.5.0 will depend on a new SOAP.py implementation

The 0.5.0 will drop my own SOAP.py module (it was a patched 0.9.7 version of SOAP.py), as that together with the docutils stuff breaks PyDS. There was a weird memory leak I was hunting for ages which I could track down due to SOAP client calls and docutils reST rendering. Really weird, you don't want to know about the details.

So I switched the stuff in 0.5.0 to use SOAPpy 0.10.1 instead. If you update to the latest CVS version, you must install SOAPpy!

You can get it at http://pywebsvcs.sourceforge.net/

As always, I made a hacked-up debian package for it and added it's dependency to the debian stuff in PyDS.

The installation description in CVS is updated already, the installations instructions on the web will follow shortly.

Since I threw out my old patched version, the patches are mostly lost - they were about string conversions and character encodings. So if you see anything weird in SOAP stuff (especially with latin-1 chars), drop me a note so I can fix it.

RSS and bandwidth

Since this question pops up from time to time, some clarification on the issue of RSS feeds and bandwidth usage.

There are actually two methods implemented in many aggregators that help conserver bandwidth:

1. conditional get

2. skiphours

The first one is just the standard HTTP conditional get you would expect: watch out for Last-Modified or ETag headers in the RSS feed response and send them in with If-Modified-Since or If-No-Match headers on polling. If no changes are detected, you get a no-change-response.

The second one is a structure in the RSS feed that is filled with the hours (GMT) where your blog doesn't have any postings. The aggregator shouldn't try to poll at all in those hours, as it is highly unlikely that there will be something new. Usually these are the sleeping hours (if the author has something like a regular sleeping habit ).

So there are solutions against bandwidth overusage and they should be applied by aggregator writers. Most good aggregators already do. Hey, even the Python Desktop Server does .

Clientside for AccessRestrictions

I now have a preliminary tool for the Python Desktop Server to make use of the new Access Restrictions for the Python Community Server. The title links to the source code. Just put the module into ~/.PyDS/tools and restart your Python Desktop Server. You now should have a AccessRestrictions block on the right side of your desktop.

This is a preliminary tool, it will be included with 0.5.0, so if you update (or use the CVS version), be sure to remove it, as it would be loaded twice.

Update: there was a bug in the generated HTML code of index_html in the tool that prevented funcitoning in some browsers (for example opera) when trying to add locations. The linked tool is updated now.

user supplied access restrictions on community servers

The linked story is about a user supplied access restriction mechanism I implemented for the Python Community Server. This allows users to set up access restrictions and user accounts for visitors. This is done through an XML-RPC interface. The capabilities of this accessRestriction mechanism can be compared to the Location directive of Apache in combination with .htaccess files and a single password list.

Hackers and Painters

It's maybe linked otherplaces, but I just read it and really like it. I can second many of the ideas expressed in that text.

BTW: I am a hacker and photographer

New beta 0.4.23: several small enhancements

Attention: the download path and all references to developer resources (cvs web access and cvs anonymous access) and package repositories have changed from kenny.gws-online.de to simon.bofh.ms (this will be the home of the Python Desktop Server development from now on). Please update your bookmarks!

Rescanning the SCSI bus on Mac OS X or Darwin

Pictures? I give you Pictures!

The PicturesTool gives you many ways to add pictures to your weblog. The first step though is to activate it. Just go to your picture preferences and fill in the picture path to point to a directory locally on your machine. Now to add a set of pictures, just put them in there (everything that PIL can read is fine). Then go to your pictures upload page and add a title and description for any pictures you want to upload. Press the upload button. Now in your pictures index page you have all your uploaded pictures listed. To use them in a story, just call the following macro:

$pictures.imageTag(PID, (200,150))

Put the pictures ID in instead of PID and maybe use another bounding box (the example uses a 200x150 pixel bounding box - pictures are always rescaled to fit the bounding box with preserving the aspect ratio of the original picture).

Another option to adding them to postings or stories would be to publish them as a picture blog. To do this, you just need to go into each picture that should be published in your picture blog and set the publish property to yes.

That's all to it, have fun!