Building a Retro Web Portal
14 Years since I made this blog.
Lot has changed….
I spent 2015-2020 with a minimum web presence.
Go back into communities and see how everything seems foreign… Things that seemed insane in 2012 commonplace in 2020, communities that ban based on 5 year old blog posts, purity tests just to ask for help on software, decreasing software quality, virtual mobs trying to get people fired from their jobs… All of that happened to me in the span of a month.
Fun stuff…
I changed jobs. Moved. Dealt with a boss that tried to make me break federal labor law….
Changed jobs again. Become a licensed Professional Engineer. Got married. Moved again. Broke my rule of no social media, made an account on X, remembering something that I said in 2012 that hell would freeze over before I used Twitter.
Maybe Hell did freeze over and now we’re here, looking back on what has changed over a decade and then some.
But this isn’t about ranting. For the first time in a while, I approached something on my to do list.
A new personal homepage for my home network.
In the 2000’s, I was rather fond of custom homepage portals like iGoogle and My Yahoo. When I switched to using the Google ecosystem after GMail made its hayday, iGoogle was their homepage portal. For those who weren’t around then or didn’t bother using it, it looked something like this:
When iGoogle was discontinued, I put down on my to do list that I wanted my own homebrew replacement. I made a stopgap measure of a basic HTML page that had my frequently accessed links all in easy reach. It stayed that way for as long as I remember now.
AI comes around. I was ‘meh’ to the whole thing, until I had it generate questions for my PE exam to cover the fact nobody wrote review books on it in recent history. After using it to correct a long standing CSS issue with my site (who would have thought just two lines would break wrapping if you had long amounts of content?) I finally got more motivation to go do other things. One thing I hate to say but needs to be said, most online help communities are now full of gatekeepers who kick out anybody with a slightly differing opinion, people who might need to get mental help running the show for some, and people from 3rd world countries trying to get unpaid help for something they’re personally being paid to do (and quite rude). AI definitely helped me bridge the gap on simple questions you couldn’t get people to give helpful answers on. We’ll come back to the AI part later.
Now I didn’t want an exact clone of iGoogle or My Yahoo. I wanted something that felt slightly more modern, but had all scripting done server side, as old sites did, and as little JavaScript as I could get away with… Mainly because I don’t know JavaScript.
I do however know PHP, in fact the companion website back in 2004 had a unreleased CMS written in PHP (I want to say version 4 or 5, but can’t remember) that never got past an initial testing phase. The test site was plagued by motivational issues that me and a few friends had working on it, also not counting a guy who now looking back at, was definitely bipolar. That guy the moment he saw our test site go live, would somehow hack into it using a server side exploit (it was free hosting after all), and would replace all the images with goatse. After a few months of dealing with that, this site was ultimately shelved from its original goal and the companion site to this blog just remained a static site.
That was called StrangeCMS. I’m not sure if there’s another software project out there with that name, but the intent was to make a public domain CMS suite that ran on PHP that anybody could remix and use. We never intended to make money from anything we did, and it was going to just be for fun. The site design was going to look similar to that of php-nuke back in the day and would have a modular based widget system that end users could use to switch out their personal sites within the broader site as a whole, after all the original intent of my site was to have a place for people to host their random creative creations.
Fast forward to 2025. I just took the PE exam for the first time (and failed) and wanted to take my mind off of things while waiting for exam results. I looked at the code for StrangeCMS, which I had two versions of laying around still, not entirely in one piece, and tried to piece together both versions to make something workable. As a full disclaimer, outside of writing VBA and small powershell scripts at my job (and prior jobs), I haven’t done any actual programming since college, in fact I wanted nothing to do with it for personal reasons. This caused a handicap for me to do anything involving programming, in fact I had to relearn some of PHP just to read my old code.
So between whatever version StrangeCMS was written against and now, several versions of PHP have come and gone and just throwing it on a local webserver without any of the sql related code threw tons of errors with the PHP interpreter. Not only that but the CSS for the CMS software was absolute dogshit and I just decided to toss anything user facing and replace it with a simple 3 column system with no sidebars. The static site test without php code worked just fine.
Now this is where AI came in. For laughs, I threw some of my php code into a chatbot and asked it to tell me what was wrong with it. It found everything wrong that I was seeing in my servers error log and gave some suggestions for fixes. Not trusting those, I spent a weekend reading through as much documentation and came up with some of my own solutions. They were rather close to what I saw from ChatGPT.
After some time playing with the scripting I finally had a basic homepage with a RSS feed reader… that couldn’t process Slashdot. At least the weather conditions applet pulled the correct information from the NWS public API.
At this time I got sidetracked, playing with CSS and also getting the weather applet to work, I stole the logo from my linkspage and got something that looked a bit better. An example below is one with the “GeoCities” theme I made for laughs.
Widgets still looked like trash and I spent some more time working on it, but eventually got something a bit better.
Unfortunately, I still couldn’t get Slashdot’s RSS feed parsed properly. Looking at the actual RSS, I realized it wasn’t modern RSS and that’s why it was totally fucked up. That was a rather easy fix and then the project was shelved until 2026.
Fast forward to 2026. In my mental stupidity the year before, I made a crap ton of themes that could be toggled, but each site page design change required me to edit 6 CSS files. I really jumped the gun on that part. I overhauled the portal at this time to make it look closer to iGoogle and then used AI to merge the changes across all the CSS style sheets. Automation really came in handy for that.
This is the “Yohoo” theme. I’m still working out some issues with widgets and whatnot and will probably make a page on the site shortly to download it in its alpha state. I don’t have interest at this time doing it on Github or anything like that as I’m making it strictly for myself. The code will be 100% public domain, as I really don’t care what people use it for.
In terms of how far back in time you can use this site on, so far it works as far back as Firefox 2.0 (and kinda works in IE6, but not Firefox 0.7.3) in terms of browsers, making it usable on at least Windows 2000 if you have a old machine laying around that you want to go nostalgic on. As I’m targeting a slightly modernish 00’s web design I don’t think I’ll make it any further backwards compatible.
This is what it currently looks like, as shown on a Windows 2000 virtual machine I have. I think it’s been a good side project for myself and hopefully somebody else might enjoy it too when I release it.
