There will not be any summer of code with Xojo. They have a closed source product and they want to leave it closed source. They can’t hire C++ guys, compiler guys and pltform guys to improve it while it would cost too much money. There product is good for low level cross platform desktop. Web? No Idea. Mobile? No Idea. In simple, Apps like for example Infor (ERP System) are completely written in Java on Server side and html/javascript on Client Side. Running like hell. That’s it. And everybody could write this kind of apps with Java. So for me it is simple cause I have my platform and my platform decision.
1 Like
we just call him “the genius”
2 Likes
It has nothing to do with it. There is a small difference. It is a real small company. You can’t have all this positions in a one man show. So this company is driven by one person with two hands full of employees. Nothing bad on it but it is a complex thing.
We have here no chance for changes while this positions will never really exist. When Mr. Perlman retires what then? It is a bit like looking brealing bad. Ups and downs on all edges. So Xojo is trying to get through this times and stull exists. There is no Million User Market for a company like them
this is exactly what I am saying for years… Xojo will encounter same fate like Powerbasic many years ago… a tech dictator taking “his baby” with him, ignoring his responsibility towards others, who trusted and believed what he preached all the years.
Maybe the sweet spot is already missed to turn around the rudder… open source xojo… broaden the developer foundation for it… fix bugs and stuff, axe maybe functions (Android, iOS) do less marketing… this would make me consider to come back.
What he is doing is madness… trying to keep selling what’s left while increasing the bug bow wave with every version… until final collapse, a steady, slowly one…
you might be surprised how well it does with more popular languages.
i have had terrible results with less popular languages (although i didn’t feed hundreds of thousands of lines of code, and reference manuals into them to build a specific model for any of those languages, unlike some other folks i know), but with more popular languages, the results have been surprisingly helpful, and have dramatically improved my productivity.
Not sure I get that. Can’t really just use your brains since your brain doesn’t have all the docs in it.
Your brain does needs the docs plus a search engine of some sort… AI is a bit like the Google “I’m feeling lucky” button. Like @mikey said, it works great for popular languages. It’s not perfect like a Google search isn’t perfect. Just like search engine results, you have to understand the results. What’s interesting about AI is that you can ask question after question to help your understanding. With Google search you have to check page after page for multiple results and in your head align them…
AI saves me a ton of time. Just yesterday, I wanted to do something in NGINX. I had no clue. So I pasted in my NGINX conf, asked my question, and after some back an forth, I had a solution. 
Trying to do that with a search engine would have taking much longer.
2 Likes
i’ve seen a couple of what i think are really good descriptions of where we are, today:
- it’s a quick replacement for RTFM
- it’s code-complete on steroids
i stumbled into this forum, a couple of weeks ago, as i was starting to search for a replacement for livecode for use at my companies. in that time i have:
- watched a ten-hour video crash-course on dart
- written three dart+flutter apps. one was a simple weather app, one was a sample online store interface, and one was for helping me chase customers who regularly run afoul of payment terms.
now, on the last point: “written”. i barely wrote anything. i prompted my way through all three of them, and went on deep dives into the weeds, when copilot was not helpful. if i was trying to learn this language and this framework, before a few weeks ago, i already would have quit and gone in a different direction.
the llm’s aren’t even close to being perfect, or, in some cases, even close to being really good, but they’re definitely a lot better than i hoped.
1 Like
TBH the pool of people who meet the criteria of very devoted Xojo developer with C/C++ skills is pretty small
2 Likes
They may not be perfect but they may be the thing that shoves you in the right direction
Even that is helpful
Although I still find a decent search of SO can also be very helpful
1 Like
Agreed, there are not too many Xojo developers with C/C++ skills.
One of the reasons why I needed to fine-tune my C/C++ skills with Xojo, is to expand the current language, create electronic drivers that don’t exist, increase graphic speed, etc. After a while, it was just easier to program in C/C++ than to make the plugin code compatible with Xojo.
Xojo is good for a 24-hour skunk-works project (make something really quick). When I wanted performance, drivers, and increased options for controls, then C/C++ coding was a necessity.
The reason why I wrote the >1200 page Declare book was because of the requests that I was receiving from clients that didn’t exist in Xojo. In my opinion, programming Xojo Windows Desktop has less options and stability, and Xojo Mac Desktop seemed to be more reliable.
1 Like
I correct this for you: There are not too many Xojo developers at all.
10 years ago it was already an issue with customers. Today it is a death penalty for a project if its tools are not open source, at least for my projects in my bubble.
2 Likes
Not being open source may be an issue but I think there are even less drastic things they could address that would be hugely beneficial
A cmd line compiler making integration in various work flows & automation easier would be a decent step
But I suspect its relegated to the “oh no average user will care about that” pile
1 Like
True… if I may give a nice example… wouldn’t work with Xojo…
This is a small crosscompile Shell-Script of my Tool Clipboard Auditor. It crosscompiles executables for both Linux and Windows, creates a tarball for Linux, SHA1 checksums for Downloads and an RECENT.md File for current downloads. (Link)
#!/bin/bash
FILENAME=clipboard-auditor
LOCAL_DIR=~/projects/$FILENAME
DLINK="https://codeberg.org/tomas-jakobs/clipboard-auditor/raw/branch/main/bin"
cd $LOCAL_DIR/src
# Compile and move Windows Executable
fbc -O 3 -s gui -strip -target x86_64-w64-mingw32 $FILENAME.bas
sha1sum $FILENAME.exe > $LOCAL_DIR/bin/$FILENAME.exe.sha1
mv $FILENAME.exe $LOCAL_DIR/bin
# Compile, tar and move Linux Executable
fbc -O 3 -s gui -exx -w all $FILENAME.bas
sha1sum $FILENAME > $FILENAME.sha1
tar -cvzf $FILENAME.tar.gz $FILENAME $FILENAME.sha1
sha1sum $FILENAME.tar.gz > $LOCAL_DIR/bin/$FILENAME.tar.gz.sha1
mv $FILENAME.tar.gz $LOCAL_DIR/bin
rm $FILENAME $FILENAME.sha1
cat <<EOF> $LOCAL_DIR/LATEST.md
## Latest Downloads (unstable)
[Windows 64 Bit $FILENAME.exe]($DLINK/$FILENAME.exe)
SHA1 : $(awk '{print $1}' $LOCAL_DIR/bin/$FILENAME.exe.sha1)
[Linux 64 Bit $FILENAME.tar.gz]($DLINK/$FILENAME.tar.gz)
SHA1 : $(awk '{print $1}' $LOCAL_DIR/bin/$FILENAME.tar.gz.sha1)
EOF
Damn I wish @Stretch327 would shine chime in with his experiences trying to automate the Xojo IDE 
its one of those things that I suspect most hobbyists wont EVER care about but any enterprise trying to work in decent sized teams and have a CI setup cares immensely about
And NOT providing it very clearly says “we dont care about that class of user”
So many actions have that feel
1 Like
Hence why we’re trying to get the hell out of Dodge.
3 Likes
I not only got out of Dodge, but Chevy and Ford as well.
You should check out Pieces AI.
1 Like
ok, github copilot just surprised me, again.
in app #3, i’m using a date picker library. i wanted to change the behavior of the library, specifically how date ranges are managed, and i wanted to use my modified library but not lose track of the original, so future updates in that library don’t leave me behind.
then i also wanted to customize the calls to this customized version of the library, to get this behavior exactly the way i wanted.
copilot:
- updated the library
- monkey patched it
- threw the new calling code at my app.
- proposed a PR to submit to the author of the library to turn the patch into a feature, because nothing is more annoying when i’m trying to plan flights than the way the date range widget works.
first try success.
1 Like
If Xojo was coded right, they wouldn’t need any C/C++ skills as it would all be done in native Xojo. That’s part of the problem. If they moved to doing everything with Xojo and declares, they could cut out their fat framework too, which would make app sizes much more comparable to the competition.
I actually wanted to fix up Xojo’s Mac framework, but Xojo didn’t want my help.
2 Likes
Quite honestly there ARE things they cannot do in Xojo that CAN be done in C/C++
Like preemptive threads
There are others
1 Like