Reddit is US centric, fragmented for the rest
Source: Reddit Users by Country 2026
107% of Canadians use it ?
WOW !
They need to count unique IP’s or something else
You can read their methodology in the fine-print. It’s like a “visitor”, but not short term, if same IP visits many times in a period, typically one month, it counts one. So same IP of someone that visits the site frequently can count as 1 to 12 “users” in a year. 107% shows lots of visits, but peanuts globally (5.01%)
107% of Canadians is just a nonsensical statement regardless of an obviously flawed methodology
Back around 1995, I did a brief stint for a marketing company in Phoenix that based a lot of its work on data that I demonstrated to the owner of the company was dodgy at best – rather reminiscent of the methodology of this “study”. He dismissed my findings with this gem:
“The data is the data. It has its own integrity.”
It was my first concrete introduction to the extent to which some people are willing to ground their assumptions in something other than reality, in the service of motivated reasoning.
Did you move over any apps from Xojo already? I’m playing around with Objo and the syntax is very similar, but it’s hard to straight up copy my code/methods since I also have lots of old API1 code spread in there. While simple methods can be moved over, I’m dreading the larger apps… How do you convert your apps without investing months of work?
I have done it in 2 months time start to nearly finish. I didn’t do straight cut and paste and do rewrite with streamlining and work even better than before..
I have tested with 1 million records and and only take a few seconds instead of hours on my old system.
my application is a collection system that have images, reports and about 20 windows and also a report writer and export writer.
I am not using Xojo any longer for more than 5 years. So I start with empty projects because I have no Xojo projects to convert.
Kind regards,
Chris
Xojo’s attitude and pricing will end them. It’s all Geoff’s choices. It’s a matter of time…
I had a Xojo app that I asked https://pi.dev to convert to flutter. It ran and I tried it and holy crap the app worked. I didn’t recall providing the Xojo source but it figured it all out from the product web page info.
Then I had a Xojo console app that I needed to convert but I had an xml project. That took even less time. Xojo http listener to php in minutes.
If Xojo has you captured, it sooooo easy to escape if you embrace AI. And you can have as much control as you want.
And this works for FileMaker too, who are in my opinion greedy like Xojo. FileMaker can export to XML and had an XML clipboard format. So AI can use those text formats to help you escape too.
Moving Xanadu to PHP was the best thing I ever did both FileMaker and Xojo changed the rules and I stopped playing. AI can take a xojo or filemaker or whatever cand convert it to open source platforms. I’ve exported my FileMaker databases to xml. My next step is to define Xanadu bundles to absorb FileMaker databases as bundles with shared table like Contacts, Addresses, Comms…
I was making progress pre-AI but now the process is so much faster.
I agree. I’ve easily translated two of my Xojo projects (without AI) to both Objo and Buoy as an exercise to learn each tool. So I can keep using a BASIC-like language and not depend on Xojo.
I’ve also been programming in Swift for a while, but recently started to use AI to ensure that my apps are fully compatible with the latest macOS and iPadOS standards.
I also used AI to convert an 11 year-old Objective-C iOS program of mine to modern Swift/SwiftUI for macOS, iOS and iPadOS in just a few hours.
Finally, I used AI to write a new app from scratch. It replaced one of my old Xojo apps with a new Swift app whose capabilities far exceed what the Xojo app did.
Most AI agents I tried could recognise my Xojo code (although my older apps are an API1 / API2 mess), but Objo is still too new to have these agents help me with conversion. I tried copy/paste but it’s a hell of a job to manually make all these small changes needed to get everything to Objo syntax.
At this moment I’m still working a lot with Xojo and although I’m using an older version from 2022 due to the current high pricing of licenses, it still fits my needs. Besides that: I have a couple of apps that I’ve been developing for 20+ years now, so that’s A LOT of code with most of it still being API1.
It’s worth trying Objo’s AI assistant linked to an AI API (either paid or local) or providing a frontier model like Claude with Objo’s MCP connection and have AI help you port. Objo Studio has a lot of tooling to help AI tools work with Objo.
AI Assistant in Objo Studio is brilliant!! have been using since I start development on objo on 8th August. I am converting my Xojo application and have most of the feature working..
I mostly rewrite the code instead of cutting and pasting. I make the new Objo version so much faster, so much easier.. and I am so happy to do it.
I am using Ollama with Qwen 3.8 for AI Assistant
On which hardware? It’s always frustrating when people using local AI don’t mention what hardware they are using to put their experience into context.
I am on macos 27.0.1 on mac mini m4 pro with 48gb memory
Qwen 3.8 27B mlx on an M2 Mac Studio with 64G of RAM.
I could probably squeak by on half that, but I think @trishaduke’s 48G is a practical minimum for anything resembling a general purpose coding assistant.
OMG no wonder RAM prices are exploding 
Garry is right, I’ve ported and improved one of my applications written in 4D using Codex and MCP Server in Objo Studio, all on $20 a month plan, in less than ten days. Then I added features I always wanted to add in another 10 days. Let’s put it this way: I wrote application that uses 40 tables in SQLite database without writing one line of SQL on my own. For Objo part, I added a class for each table manually and then let Codex do the rest, sometimes I started designing windows, but now I let Objo design them and I fine tune them. Let’s say that I didn’t write more than 10 lines of Objo Basic code.
If you are already paying for Claude or ChatGPT, try MCP Server feature in Objo Studio first. It is working better for me than using AI Assistant with OpenRouter and free model at that end, maybe I am just using the wrong model.
For my local AI I am spoilt - I have two DGX Sparks with 256GB memory so I run GLM 5.3 Flash locally. A lot of my users have had good success with models in the 27 - 30 billion parameter size for use in the AI Assistant in Objo. Of course if you want more intelligent, use a provider like OpenRouter and pick from one of the many excellent open weight models out there (DeepSeek v4.1 Flash, GLM 5.3, MIMO v2.6 Flash, etc).
Each? 