Michael Jay Lissner

New Version of the Site is Now Live

2014-08-27T00:00:00-07:00

I have big news today for the small world of people who read my blog regularly: A new version of the site is now live and the old version shall die a quick death.

Version 4 was pretty nice though, while it lasted:

Other old versions of the site still available!

The improvements

This new version comes with some big improvements that I’m quite pleased with:

If you find typos in a blog post, you can edit them on Github and I can easily integrate your changes. Check out the link on the right to edit the typos in this very page. (I’ve left a few conspicuous ones as a treasure hunt for the reader!)
The site is now much faster and can handle immense traffic without a hitch, thanks to being hosted by Github Pages. The previous version would have occasional hiccups during times of high traffic – something that’s really quite untenable.
Comments are now moved to Disqus, though unfortunately old comments have not made the jump to the new version of the blog. Comments are collapsed by default so the scrollbar actually represents the length of a highly-commented post.
The site now looks bad-ass. Regardless of whether you’re on a phone, tablet computer, or what-have-you, it’s going to look good.
All content has been categorized as well as tagged, as you can see in the sidebar. There are Atom feeds for each.
The homepage has a new design that focuses on my projects and bio, and then has recent posts below that.
Long articles like this one get an automatic table of contents on the left.
The site is now optimized for speed dial in Opera and to be made into apps on mobile phones and tablets. For example, if you’re reading this on Android Chrome, you can simply click “Add to homescreen” in the ⋮ thing and you’ll be all set.
Security is now invincible: No more webserver to update, no more database, no more outdated Drupal. It’s basically impossible to hack the new site. I’ve also added my PGP key to the contact page, for those interested.
The entire site is now static and doesn’t require that I pay for or maintain a server or database. Bonus!

So those are the high-level changes you can see as of now. If you’re interested in the technical nitty-gritty, read on.

The Tech

The original motivation to rebuild the site came when the old version kept overwhelming the server that was running it and requiring that I step in to make it work again. And if that weren’t annoying enough, I have been paying for that server for the past several years, which just seems a bit silly for a simple blog like this one.

The solution? A so-called Static Site Generator or SSG. With one of these, the paradigm for your site totally changes. Instead of having a dynamic site that loads every time somebody visits the page or makes a comment, you generate the entire website on your laptop (this takes about 30 seconds), creating static HTML, and then push that to some cloud provider of choice (in my case, I use Github pages for this because it’s free and easy).

There are about 300 SSGs right now and the one I eventually landed on was Pelican due to it being written in a language I knew (Python), and due to it having lots of good themes and plugins. I briefly tried to make a switch to Hugo instead because it’s written in Go and is much faster at generating content, but the documentation for Hugo isn’t very good yet, and it doesn’t support basic pagination, which is something of a showstopper.

Switching to a SSG from Drupal

Switching from Drupal was pretty awful and took a lot of effort — days of it! The goal was to get all of my posts exported from Drupal, convert them all to markdown, and to get them all live on Github pages. Let’s go through this process together.

Exporting from Drupal

This step of the puzzle was, shall we say, a pain. Nobody has yet made a Drupal to Pelican converter, so I had to do it myself. The script that I wrote dug directly into Drupal’s database, pulled out the contents and converted them to a format that Hugo could understand. At the time I thought Hugo would be the SSG for me, but later I switched to Pelican, and had to write another script to make the conversion from Hugo to Pelican.

Problems with Drupal

This was a good start, but Drupal has a few funny conventions. One is that it allows files to be “attached” to blog posts. Most blogs don’t do this (Pelican and Hugo included), so I had to go through all of the items that I attached to Drupal posts and convert them to inline links instead. This took a while.

Another problem I ran into is that the posts themselves were written directly in HTML, which makes them kind of awful, and not very portable between blog engines. Content for Hugo or Pelican should be written in Markdown, so I began making this conversion to the 200+ posts on the site. In general, the process for this was to find a post and begin cleaning it up. If I encountered something that a computer could reliably fix across all the posts (for example, <i> can be converted to * and <strong> to **), I wrote a little script to do so. In the end, this took a lot of time, but I now have a collection of a few hundred nicely-formed markdown files that power the blog.

Moving to Github Pages

With all of the content converted properly, the remaining step was to get the project live on Github. I found this process confusing, but the process is basically this:

You need to take the output file from Pelican and put it into a Git branch called gh-pages. To do this with Pelican is remarkably easy, as there is a simple command you can run: make github. Run that, and you’ll be all set, with the content pushed and everything.
You need a file named CNAME that simply contains the domain of your website. This is easy in theory — it’s just a plaintext file — but in practice it is difficult because you need the file to be created by the make github command mentioned above. To do that add the CNAME file to a directory at content/extra/CNAME and then add the following to your pelican configuration file:
```
EXTRA_PATH_METADATA = {'extra/CNAME': {'path': 'CNAME'}, }
STATIC_PATHS.append('extra/CNAME')
```
Do that, and the file will get copied over whenever you run make github.

If you’ve done this correctly, you’ll see evidence of such in the repository’s settings page on Github, where it will tell you the domain in the CNAME.

You need to configure your DNS provider to point your domain to Github. This varies by provider, but I can tell you that your final version should look something like this:

↪ dig www.michaeljaylissner.com +nostats +nocomments +nocmd

; <<>> DiG 9.9.5-3-Ubuntu <<>> www.michaeljaylissner.com +nostats +nocomments +nocmd
;; global options: +cmd
;www.michaeljaylissner.com. IN  A
www.michaeljaylissner.com. 3600 IN  CNAME   mlissner.github.io.
mlissner.github.io. 3600    IN  CNAME   github.map.fastly.net.
github.map.fastly.net.  2   IN  A   199.27.79.133
fastly.net.     66087   IN  NS  ns4.p04.dynect.net.
fastly.net.     66087   IN  NS  ns3.p04.dynect.net.
fastly.net.     66087   IN  NS  ns2.p04.dynect.net.
fastly.net.     66087   IN  NS  ns1.p04.dynect.net.

Final Words

This is been a much larger undertaking than I expected, with tons of corner cases that I wanted to fix before releasing a new version of the site. In the end though, this has been a good investment that I can expect to keep the site going for the next five to ten years.

I hope you enjoy the new look and new features.

Our New Zealand Blog

2013-11-27T10:19:46-08:00

If you seek our blog about our journey on the Te Araroa trail, it lies down this path.

Privatizing the Twitter API Feed

2009-03-20T12:34:31-07:00

UPDATE: Check the comments for a version with caching.

A friend of mine recently had a rather unfortunate event involving her twitter public timeline, so I thought the time had come to make mine private, more or less.

As a result, I needed to update the code that pulls my most recent Twitter posts into the left hand column so that it would authenticate using the Twitter API. Here’s the new code - it ain’t pretty, but it works:

<?php

    // Your twitter username & password.
    $username = "YOUR_USERNAME";
    $password = "YOUR_TWITTER_PASSWORD";

    //Concatenate the username and password
    $userpass = $username . ":" . $password;

    //Make up the feed URL
    $feed = "http://twitter.com/statuses/user_timeline.atom?count=1";

    //A function to parse the atom feed and pull out the useful info.
    function parse_feed($feed, $username) {
      $stepOne = explode("<content type=\"html\">", $feed);
      $stepTwo = explode("</content>", $stepOne[1]);

      $tweet = $stepTwo[0];
      $tweet = str_replace("&lt;", "<", $tweet);
      $tweet = str_replace("&gt;", ">", $tweet);
      $tweet = str_replace($username . ":", "", $tweet); 
      return $tweet;
    }

    //Create a curl object, give it the feed and authentication
    $curl_handle=curl_init();
    curl_setopt($curl_handle,CURLOPT_URL, $feed);
    curl_setopt($curl_handle,CURLOPT_USERPWD, $userpass);

    //Return the result, don't print it.
    curl_setopt($curl_handle,CURLOPT_RETURNTRANSFER, 1);

    //Make the connection, set the variable, close the connection.
    $twitterFeed = curl_exec($curl_handle);
    curl_close($curl_handle);

    //Echo the parsed feed. Done.
    echo parse_feed($twitterFeed, $username);

?>

One dependency is the php-curl library, and after you install that, apache2 will want a restart.

Twitter (and Facebook) Integrated

2009-01-25T13:03:03-08:00

I upgraded the site a bit today by adding my Twitter/Facebook feed to left-hand sidebar. To a teenager in Colorado I am indebted for this script. Jeez, they just get younger and younger.

Let me know if you catch any bugginess.

Marvelous RSS, Marvelous Google

2008-01-27T15:36:34-08:00

I’ve been working on the site a fair amount lately, and have added a couple of new features. The first one is a set of those ubiquitous bookmarking buttons for Technorati, del.icio.us, etc. that you should be seeing if you are reading this online. I’ve been trying to optimize this blog for a bit, and I figured I probably need these sooner or later.

The other thing I added today is the “Subscribe” block, which for the moment is listed on the left under the recent music. I did a little playing around in the heart of Drupal, and modified this block so that it has text links for RSS and for an email service I just discovered from Google.

Taxonomy Features Are Go

2008-01-13T16:56:44-08:00

I have been rather busy since starting this blog many months ago, so I never got to finish rounding out its features. One that I have been wanting to figure out and to apply was Drupal’s taxonomy feature, which allows you to categorize your content as you write it.

After I turned it on today, I had to make up about 40 terms categorizing the content that I have written thus far, and I am now in the process of quickly going through and applying these tags to my old entries. What’s cool about having this feature turned on is that in the future, say, when I have 80 gazillion entries about various topics, if you are a reader of the blog, you can easily find entries matching your interests. Assuming, of course, that I write 80 gazillion things.

Old Versions of Site are Up

2007-10-11T19:04:42-07:00

I was working on getting the old versions of the site up and going for history’s sake, and I have more or less accomplished the task, though I learned a very important lesson in the process.

For one, I learned that if you want to host additional custom directories on your drupal site, it’s a piece of cake: Just make the directories in your drupal directory, and you’re done. For example, my drupal directory is at /usr/share/drupal-5.1, so to host a page at michaeljaylissner.com/oldsite, I just made a directory at /usr/share/drupal-5.1/oldsite, put stuff in it, and was done.

The more important lesson was that when hand-coding HTML, as I did in the previous version of the site, relative links such as /assets/picture1.jpg are NOT the same as relative links such as assets/picture1.jpg. Both work, but only one is actually relative to the current page (the latter one). The former one may as well have been written michaeljaylissner.com/assets/picture1.jpg, because that’s what that first slash means. Duh.

Somehow, I never ran into this problem before now. Anyway, if you take the time to check out the old site (which I don’t really recommend), you’ll find many broken links. This is why. The other project I have been working on is getting all of my pictures from the old site set up on this site in a place where registered users can find them. This is a work in progress, but if you’re a registered user, you might be interested to see them, if you haven’t already.

This is it. Drupal systems are go.

2007-09-12T21:46:59-07:00

Well, I can’t say I have had a serious blog before, so with this very sentence, I am forging new ground for myself, and through the strange magic of the Internet, for everybody else in the world (except China, who I will now block from this blog by writing the following word: “democracy”).

There’s a pressure here of course not to have any typos and to make this post significant and poignant, but if I know me - and I think I do - I shouldn’t set high bars like that for myself. I should just enjoy the medium, allow mistakes early and often, and proceed with the reason I am here to write today, which, according to the title of this post is to discuss the software of this very blog.

It’s an interesting thing setting up a blog in your house. For some reason, it’s really bloody complicated. First, you’ve got to learn how your house is networked, then how the Internet is networked. Once you think you’ve got that figured out, you have to get a faster Internet connection (cable as opposed to DSL), and if you are feeling spry, perhaps a static IP. I’m not feeling spry, and so, dyndns.com is my friend.

The next step once the networking is figured out is to figure out the software and hardware sides of things. This is not an easy task either. In the end, I have set up two computers to run this here blog. One to run Drupal, Apache, etc., and the other to do the mail serving. For the CMS, I have chosen Drupal because it seems to be all the rage, and because it has a stupid song that got caught in my head for far too long (see youtube for details). Drupal is of course running on GNU+Linux, Apache, PHP, and MySQL. As for the mail server, it would be running Zimbra, but at the moment it is crashed. I can’t make it stop overheating, and it’s rather annoying.

Anyway, once that’s up and going, everything should be a go, and I will be a proud Internet host. Only one question truly remains: Is my IP on an anti-spam blacklist? We shall soon find out.