The photo in the title is one I took while playing around with my new 50mm f/1.8 Nikkor lens. With the right adapter the lens can be mounted in reverse and used as a pretty decent macro lens. I didn’t have the adapter so I just held the lens up to the camera body instead and started goofing around. When I saw that shot I immediately thought about using it on the site, and it seems to fit perfectly IMHO. Since then I’ve bought the reverse adapter ring (they’re dirt cheap on ebay) so I’ll be sharing some more of those shots in the near future on my Flickr page.

The Flickr page is another “news” item. I’ve had a Flickr account for a long time now but never really used it because of various limitations in Flickr. But now I realize that I just don’t want to deal with maintaining my own photo hosting software. Flickr is reasonably priced, has unlimited storage/bandwidth, and gets my shots into an environment where real photographers might actually look at them and give feedback.

Finally, I thought I should mention that I upgraded across not 1, not 2, but 3 WordPress versions (from 2.0.something to 2.3.2) in a matter of minutes thanks to subversion. My installation is a subversion checkout of the WP developers’ repository, so all I needed to do was svn switch to the 2.3.2 tag to upgrade. I was able to use svn diff to find the minor tweaks I’d made to some of the WP files and svn revert them once I determined that they weren’t needed anymore. This is definitely the way to maintain a website. Happily, my theme only needed one minor adjustment to work in 2.3.

As always, editing CSS has left me annoyed. I know that it’s important for CSS to be easy to parse and apply, but couldn’t we have simple variables or even functions? My CSS is liberally sprinkled with stuff like background-color: #f60; that says nothing about what the design intent of #f60 is. This from the organization that stresses separation of semantics from presentation? Imagine instead that I could do something like:

let border_highlight = #f60
let button_hover = #f60
...
.somebox { border: 1px solid border_highlight }
.somebutton:hover { background-color: button_hover }

Now if I want to change the border highlight but not the hover color it’s simple, and there’s no worries about accidental search/replace errors. The argument for functions is similar. If you want to define a bunch of related classes multiple times but with slightly different parameters, you end up defining one set, then overriding values to define the second, again for the third, and so on. It would be much cleaner IMHO to be able to define a function to generate these classes and then invoke the function several times. Possibly not a big win in terms of lines of code, but I think it would be easier to maintain.

It is, of course, possible to use the C pre-processor cpp with CSS, so that might be a way for me to get my wish, but I’m reluctant to add a compilation step into my already awkward web-development process. (Speaking of which, css3.info, MacFuse, MacFusion and SSHFS rock!)

So anyway, enjoy the bling, variables are good, functions are great.

The proper fix was slightly tricky, and it took a lot of searching to find help. I was trying to fix it in an .htaccess file, but the DirectoryMatch tag I needed isn’t allowed in .htaccess files for some reason. So in the end, here’s what I added to my httpd.conf file:

<DirectoryMatch "^/.*/\.svn/">
  Order deny,allow
  Deny from all
</DirectoryMatch>

I’m posting this in the ScriptExport category so that those of you who subscribe to just the SE RSS feed will get the message as well. I’ve added some redirects so that the old URLs will continue to work, but you should probably switch over to the new ones. Here are the new links:

• The main blog feed
• The ScriptExport AppCast feed

BTW, to get an RSS feed of any category, just append “/feed/” to the URL you get by clicking on the category in the sidebar.

If you're anything like me, you have a list of a few websites that you check every day. I call them my “daily fix”. Every day I fire up my browser and go to each one to see what's new. This is fine for a handful of sites that are updated daily, but there's a lot of interesting stuff out there on the internet. Daily browsing doesn't scale well if you want to keep up with dozens or even hundreds of websites. It also doesn't work very well if you want to follow a website that's only updated once or twice a month because you'll quickly lose the motivation to check in on it. RSS is a web technology that can help solve these problems.

What's RSS? To oversimplify, it's a format that a website can use to provide a compact digest of itself, called an RSS feed. For example, a news site might provide a list of headlines along with the first paragraph of each story. A photo album might provide a list of albums with a highlight photo from each. You'll often find a link to the RSS feed of a website cryptically denoted by a little orange box like this:


RSS


or this:


XML (RSS is an XML-based format.)

An RSS reader, or aggregator, is an application (either installed on your computer or on a website) that quickly scans tens or hundreds of RSS feeds and displays them for you in an easy-to-browse format. For example, here's a screenshot of NetNewsWire Lite, the RSS reader that I use:




On the left is a list of all the feeds I'm subscribed to. The upper-right pane is a list of “headlines” from the currently selected feed, which happens to be my photo album. The lower-right pane displays the “story” for the selected headline. As you can see (if you have good eyesight), it's easy to tell at a glance that Buy Blue has 15 new headlines but KernelTrap has none. If I were to select Buy Blue I could easily scan the headlines and decide if any of them were worth reading. Also, I will find out if my lab-mate Cristian adds any new photos to his photo album without needing to browse to it on a regular basis. The RSS reader allows me to focus on the content that I want to read without spending a long time browsing the web.

So now that I've told you what RSS is all about, I hope you'll download an RSS reader (Mozilla Thunderbird, for example) and subscribe to the feeds for my blog and photo albums!]]> http://www.n8gray.org/blog/2005/07/13/rss-how-to-find-out-when-things-change/feed/ 0 http://www.n8gray.org/blog/2005/04/23/comment-spam-and-turing-tests/ http://www.n8gray.org/blog/2005/04/23/comment-spam-and-turing-tests/#comments Sun, 24 Apr 2005 05:30:00 +0000 n8 http://www.n8gray.org/sandbox/wordpress/?p=66 So n8blog got its first round of comment spam last night. I figured this would
come up at some point, but I had hoped that the relative obscurity of PyBlosxom
and my site might keep the bogeyman at bay. But there it was. About 20 people
left comments, and golly, they all just happened to be enthusiastic supporters
of various casinos in France and Italy!

As an utterly brain-dead way of preventing this sort of thing from happening
again I've implemented a simple
Turing test
on the comment entry form. For
those of you who aren't familiar with the concept, a Turing test is is a
challenge posed to AI researchers. A person is seated at a terminal,
communicating with two entities. One is another human,
the other is an AI. The person is allowed to ask any question of either
entity, and must try to correctly identify the AI. The AI has passed the
Turing test if it is able to fool the person about 50% of the time.

My Turing test, of course, is not a real Turing test. You don't need to enter into a lengthy conversation to prove your humanity. Also, the tester is not a human, but a Python script. And the test is, shall we say, on the simple side?

By the way, a type of real Turing test was conducted recently, but the results haven't been posted yet.]]> http://www.n8gray.org/blog/2005/04/23/comment-spam-and-turing-tests/feed/ 1 http://www.n8gray.org/blog/2004/12/23/i-want-keywords/ http://www.n8gray.org/blog/2004/12/23/i-want-keywords/#comments Thu, 23 Dec 2004 09:31:00 +0000 n8 http://www.n8gray.org/sandbox/wordpress/?p=68 For those of you who haven't used a blogging tool like Blosxom or PyBlosxom, the way it works is that you put all your blog entries into a directory hierarchy that reflects the categories that you want these entries to go into. This is kind of appealing to people like me, who are used to dealing with files and directories. However, it forces you into the hierarchical mindset — everything must have exactly one place in the hierarchy.

However, it's not hard to imagine a blog post that concerns hacking, unix, and OS X. I've got these categories in separate branches of my directory tree, which makes it impossible for me to put my post in all three places without having it show up three times in the chronological view of the blog. You could imagine a solution involving adding smarts to PyBlosxom with respect to symlinks, but even this strikes me as unsatisfying.

You see, I've been spoiled by iPhoto. iPhoto allows you to assign keywords to each photo, which you can use to create “smart albums,” which are essentially frozen Boolean searches that update automatically as you tag new photos. I guess I hadn't really thought about the potential for something like this until iPhoto gave it to me, but now I want to put keywords in everything! I want to organize my blog using keywords so that entries can show up in multiple places. I want to attach keywords to music files so I can attach descriptors like “instrumental” or “sleepy” to songs and make smart playlists. It's sort of possible to do this now in iTunes, but it should be supported directly.

Actually, I should say I've been spoiled by the potential of iPhoto. Unfortunately, iPhoto's interface for using keywords is downright shoddy. You create keywords by adding them to a pop-up dialog that is extremely awkward to use because it seems to be based on some kind of bastard half-breed non-aqua list widget. You only get one big list of keywords that you can't even rearrange. Good luck if you want to create more than a dozen or so! To do it right we need folders to organize 'em. You also can't assign more than one keyword at once because there's no multiple selection. You can't navigate by keyboard, so assigning keywords becomes a laborious click-fest.

While I'm ranting about iPhoto, I should mention the album interface. Again, you simply get one long list of albums. Thankfully, you can at least rearrange these as you see fit, but once again we really need folders to organize our albums into. It's like the developers of iPhoto only really think in terms of 5 or 10 of anything: albums, keywords, even photos in older versions, which choked badly on large albums.

This brings me to a fundamental rule of interface design: Any time you're letting the user make a collection of things you should give them a way to organize it. At least you need to give them a way to make a hierarchy. Better still is to provide a keyword-based categorization system. (Can you imagine it? A keyword-based system for organizing your keywords? I think it would actually work. ]]> http://www.n8gray.org/blog/2004/12/23/i-want-keywords/feed/ 1 http://www.n8gray.org/blog/2004/06/23/migrating-to-pyblosxom/ http://www.n8gray.org/blog/2004/06/23/migrating-to-pyblosxom/#comments Wed, 23 Jun 2004 13:17:00 +0000 n8 http://www.n8gray.org/sandbox/wordpress/?p=76 Blosxom is nice, but unfortunately it's implemented in a write-only language. This is a problem, because I already have ideas for pushing it in new
directions but I'm constantly confronted by the Perl Problem.

Luckily, any time something worthwhile is written in Perl, somebody comes along and rewrites it in Python (and Ruby, and Scheme, and countless other
flavors of various weeks). Thus, we have PyBlosxom. PyBlosxom
is much younger than Blosxom, so it's not surprising that it's a little rougher
around the edges. The documentation is much worse than the already shaky
documentation for Blosxom. Even important plugins that come bundled with
PyBlosxom, like the comments plugin, fail to document basic things like which
variables they provide for use in your flavour files. The PyBlosxom folks would
be well advised to follow the Blosxom example and use an informally standardized
format for documenting plugins. Something along the lines of

1. Description
2. Configuration Parameters and Instructions
3. Variables provided
4. Examples

Also, it is repeatedly stated that PyXML is a dependency for various plugins, which is not true since
the xml module in question has been included in the Python standard library
since Python 2.0. Finally, PyBlosxom is often advertized as being compatible
with Blosxom flavours, but this is not true. PB uses a similar scheme, but the
variables you use within the flavour files are different and there are
differences in the set of files contained in a flavour.

Having said all that, I'm still happy to be switching. It's painful for me to work with a system that I can't tweak and extend, and I'd like to see how
useful I can make this blog be. With some improvements in the documentation,
PyBlosxom should be a good platform for exploring blog-space.

Here's the breakdown. These files just need to be readable by www:www:

• Blog entries (usually .txt files)
• Flavour files
• Plugins

Therefore, these files can live anywhere as long as the www user has read access. I put them in
/Users/n8gray/web/blosxom, which I made readable by all users.

There's usually only one directory that needs to be writable by the web server: The plugin state directory. You want to make sure, however, that only
you and www are able to write there. One easy way to do this is
to make the plugin state directory owned by you, but with group www. Then make
it both owner and group writable (Permissions 775, or drwxrwxr-x). I use the
directory /Users/n8gray/web/blosxom/plugins/state for plugins.

One thing to note is that none of these files are under /Users/n8gray/Sites or /Library/WebServer. Putting
them in those directories would potentially allow strangers to read raw files
or activate plugins directly from their browsers, which is probably
undesirable. We want all access to the blog content to be through the
blosxom.cgi script. Therefore, blosxom.cgi should be
the only part of the Blosxom system that lives in a
publicly-accessible area (something like ~/Sites/cgi-bin).

Finally, I don't claim that this is the best or only way to set up Blosxom under OS X. I do, however, think that this is a sensible way to set it up.
Leave a comment if you have a better idea!]]> http://www.n8gray.org/blog/2004/06/02/blosxom-permissions-and-directories-on-os-x/feed/ 1