Article on Rails with MPM worker

I’m in the process of writing a small document on using Typo with MPM worker and FastCGI. The Rails community is fixated on lighttpd for “performance” reasons. Apache httpd 2 is pretty easy to configure to get significant additional speed, but few people seem to bother.

No need for a proxy, or yet another web server soaking up all the memory on your system, you can do it all with 5 or 10 total processes. (Besides, soaking up memory is what amavisd is for.) And if you’re doing interactive things like serving web pages, you certainly do not want to be anywhere near swap anyway unless you hate the people viewing your page and want them to hate you. That people have swap 2-3x RAM size on systems with 2GB of RAM is insane. Almost as insane as letting an interactive operating system continue to create more and more page files over time as it vastly oversubscribes available memory, that would be absolutely crazy, as well as stupid and worth it’s own rant on another occasion.

In any case, I’ll have the MPM worker detail posted in the next few days. This configuration has saved me at least 100MB of allocated memory space, mostly due to 100% less mod_php. Turns out PHP as a FastCGI is smaller anyway, a nice bonus.

iTunes 5 and mt-daapd Incompatible

It doesn’t come as a huge surprise that Apple broke compatibility with mt-daapd as part of their release of iTunes 5 today. Complaints are already starting to appear from users.

Packet dumps are easy enough to generate that folks will work around this soon. It’s hard to say why Apple even bothers breaking compatibility like this since it is so trivial to work around. These sorts of intervention efforts will continue to be futile, much like the purported protection of things like the so-called broadcast flag that consumer products can ignore (or be rigged to do so). You can’t give media to an end-user and think you’ll stop them from extracting it into whatever they want to use it for. You can only make it into an attractive challenge for intelligent people who need something interesting to kill time with.

More Typo Patches

I’ve got a couple new Typo patches that I’ve made in the last week to add features that I wanted. They’re both working here on this site right now. One makes a toggle so people can limit the Atom and RSS feeds to show the top of the article, but not the extended content. The other breaks the sidebar off into its own cached element so it can be updated more frequently but independent of the article it is appearing beside.

#316 Add toggle for extended article text in RSS feeds

The default behavior of Typo is display the whole article in the RSS feed. This is nice for people who want to read all the content in their RSS reader. However, certain tricks with CSS may not render well and users of Safari’s RSS reader will have to scroll a long ways if articles are lengthy.

This patch adds a toggle to Typo’s admin interface that limits the RSS feed to showing only the top article text and not the “extended content” (if there is any). Reloading my RSS feed will result in not seeing the article full of the repairs I made to the Audi, but only the first part that provided an overview.

Some people use similar features in other blog software to drive traffic to their sites by providing only a teaser in the RSS feed. I like to think of it in an overview sense rather than trying to get more advertising traffic, but to each their own.

#345 Separating sidebar into its own cache

Typo’s current behavior is to render the the sidebar at the same time of the article and leave it in the static page cache. This is problematic for people who have things like RSS feeds or Flickr feeds in their sidebar. The only workaround so far was to sweep the whole cache regularly, forcing every page to get rebuilt.

This patch uses some JavaScript to pull a fresh copy of the sidebar as the page is loading. The cache file for the sidebar URL can be swept separately from the rest of the cache so the sidebar can be updated without forcing rebuilds of all the other pages. It also helps to guarantee consistency of the sidebar across all the pages for the site.

One of the downsides of this patch in its current state is that Google AdSense ads in the sidebar won’t load. Generally Privoxy and PithHelmet keep me from ever see these ads anyway, but I’ll have to dig deeper to figure out what is going on. Right now, it appears this is an issue with getting JavaScript within the updated content to run. I may also just contact AdSense support and see what they suggest.

There also seems to be an issue with UTF-8 not getting properly decoded in the sidebar snippet that is being downloaded. This may just be a web server configuration hangup.

Ruby-doc.org Not Working

I’ve had some frustration in the last day or two trying to access ruby-doc.org. As it turns out ruby-doc.org’s name servers don’t actually exist. But a play-by-play will make this easier to explain.

The nameservers for .org say that ruby-doc.org is served by ns1.tagbomb.com and ns2.tagbomb.com. So now we need to find ns1.tagbomb.com.

The nameservers for .com say that tagbomb.com is served by ns1.ev1servers.net and ns2.ev1servers.net. Alright, off to .net land…

The nameservers for .net say ev1servers.net is served by ns1.ev1servers.net and we can find it at 207.218.245.135. Ah ha!

When we talk to 207.218.245.135 it says that ns1.ev1servers.net is indeed at 207.218.245.135 and you’re talking to it. When we ask it about ns1.tagbomb.com (which we were told it should have an answer for) it says it does not exist.

So there we are. ns1.tagbomb.com and ns2.tagbomb.com don’t exist but are the authoritative name servers for ruby-doc.org. No authoritative name servers means no access to the domains. (Although you are okay with certain name servers that allow their idea of where ns1.tagbomb.com is to be poisoned by the .org nameservers)

I hope they fix this soon, not having access to documentation is annoying and doesn’t reflect well on the language.

Update…

It’s fixed
now. James Britt was quick about it after it was brought to his attention.