Welcome to Ordinary Times 3.0
As you may have noticed, some things have changed around here.
We are pleased as punch with the new look and once we had it together we couldn’t wait to release it. That said, it was born out of necessity and not a displeasure with the existing design which we liked and still like. There were a couple issues that the new site resolves.
Why We Did It
The first is manageability. I am not a webpage designer and the things that made the old design great were often the things that I couldn’t manage. It was mostly designed by CK Macleod, who is a WordPress programmer professionally, and the rest held over from Erik Kain. Both of these people are far better with websites than I am, which created a problem when I took over: I couldn’t adequately maintain the site. There was always a fear that something could break and I would be completely and entirely unable to fix it. It existed in a black box of code that I couldn’t read. I could run a WordPress site, but not this one. And considering I was the most qualified person to do any of this, that needed to change.
The second issue was hosting As you have picked up on, we’ve had a lot of hosting issues in recent years. The intricate nature of the WordPress installation that it was running on was such that it was taking up too much in the way of resources. Hosts either didn’t want to deal with it and wouldn’t help us or in a couple cases actually canceled or declined our business unless we were willing to get on a plan that was (or should have been) far more what our site could justify. The database on the old design was about 1.6gb in size and the new one is about 850mb. The size of the old one could be brought down, but only using tools I was afraid to use because if something went wrong I couldn’t fix it.
We’d had the idea that we would just design this site and keep it in our back pocket until such a time as we found a new host or the existing site just collapsed. An insurance policy. As I started to design it, put it together, and see what it looked like I decided I didn’t really want to wait. Partially for that reason, and partially because it meant maintaining two website designs instead of one. I asked the editorial team what they thought, and they were on board.
What We Did
This part was tricky. There was no one aspect of the previous site that was causing problems. To use FastComet (our host from December 2017 to July or so of last year) benchmarks, their top-tier shared hosting plan did not allow for databases above 1024mb and on the old design ours was roughly 1600. The bandwidth limit on their unlimited bandwidth plan was 90gb per month and a site with our traffic should have been using about 40gb and instead we were anywhere from 70-100gb, which is less about database size than site design. So if the database is a problem and the design is a problem, what do we do?
Well, new design, new database. While my PHP skills are not great, one thing I can do is databases. I even have experience with database manipulation on WordPress specifically. I had to upgrade Hit Coffee from WP 1.5 to 3.7 and there were no tools to do so. I basically had to go into the database, find the data I wanted, and put it into the database of the new site. I created a command sequence of 38 steps and it was done. So this project was going to be like that one but on a far more massive scale. But the short version is that I created a new site, I wrote the command sequences to transfer the specific data that I wanted (most notably post and comment content), and then left the rest behind. The longer version involves 213 queries accounting for all of the data we needed without bringing in unnecessary data or worse, data that might bring over some of the system complexity that I was deliberately leaving behind.
On top of all of that, I had to put together an ambitious website and replicate as many of the features as possible.
What This Means
Unless something goes dramatically wrong and I have to revert back to the old one until I fix it, it means that this is the new site! Unfortunately, some of the features of the site were inevitably lost in the process. Some we’ll get back, some we probably won’t. The good news, though, is that by having a site that is manageable we can start making improvements. Michael Cain has been tasked with bringing back as many of the old features as possible. We both have clone sites that we can do all manner of experimentation with without disrupting site operations. We may be starting from behind, but we now have the ability to move forward.
And hopefully, the new look will make visiting the site a more enjoyable experience.
A lot of people helped, but I want to call out a few people in particular. The first is the editorial team. They have just been amazing since I have taken over and have really stepped up over the last couple of weeks. Especially Andrew Donaldson, who has more or less taken point on the processing and scheduling of posts so that I could focus on the backend stuff. Second, Michael Cain has agreed to learn more about WordPress and work on the site features. While his work has not been implemented yet, that he is working on it allowed me to move forward with more basic site functionality and design. And lastly, CK Macleod who left the site a long time ago but has been there for all of the questions I had about the site’s previous design and WordPress more generally. Maybe I could have gotten through all of this without him and maybe not, but if not for him we would not be at our launch date. He also remains responsible for many of the site’s features, from Ten Second News to image handling.
And, of course, you all. While CK and Erik did an amazing job programming the site, the important thing has always been the content itself, both in the posts and the comments. The site is mostly a way to facilitate that. You made this a site worth devoting a lot of time to.
Now we see what went wrong. For instance, load times appear to be very slow. This happened on some of the test-runs and self-corrected, and in others did not self-correct. We’ll have to see what happens there. If it doesn’t self-correct, I will keep working and plugging away at it until I see why it’s happening on this version and not (most of) the other version(s).