Bonum Certa Men Certa

GitHub-Free: Why fig No Longer Supports Pygame

Article by figosdev

fig 4.9



Summary: "I don't love the idea of trying maybe Tk instead. But Tk is at least responsible enough not to keep selling themselves to an aggressive monopoly that hates software freedom."

This story is more about GitHub than it is about fig, but I use fig more than anything else I've ever written -- it's basically my reference implementation of a language for teaching.



Because of this, I like it to set an example for all my other projects; whether they follow the example or not.

Fig goes back to version 0.4, it was always based on CPython, the standard and most conventional implementation of Python -- for fig 3.x it switched to Python 3.

"I hate when languages break a lot of things, as if nothing made previously with them matters."Each stage of fig was an experiment, though I rely on it and I've tried to keep it as stable as possible. I hate when languages break a lot of things, as if nothing made previously with them matters. In an industry, that makes enough sense. For hobbyists, that really sucks sometimes.

I use fig for both serious purposes (at the moment I've got a fig program running for 48 hours on a dedicated machine, processing over a million files) and for fun. I've enjoyed tinkering with graphics since I was a kid. Fig is what I use for that.

But for dealing with various kinds of files, Python 2 suits my purposes. After years of checking on options and tricks, I did my utmost to make fig Python-3 compatible, as an experiment. Fig really leans on the way that strings are expected to work in Python. I've gone through many tutorials, run 2to3, maybe you can make a language based on fig that uses Python 3. Have fun with that.

I have fig 3.x a serious chance -- I used it almost exclusively for half a year. After that I was working on a serious project and fig 3 started mishandling a file, I ran it with Python 2 instead and it worked better for my purposes.

"I use fig for both serious purposes (at the moment I've got a fig program running for 48 hours on a dedicated machine, processing over a million files) and for fun."That's why fig jumped right ahead to 4.x, the most recent version of which (until today) was from late 2017.

If Pygame is not installed, fig falls back on escape sequences -- it draws in the term instead. It's designed specifically to have fewer than 100 commands; I created an experimental (quite nice) "figplus" to try some new things. I'm still very happy with standard fig; the main features figplus adds are extra Pygame features and some neat handling of Python dictionaries. Both allow inline Python in fig programs, so you can define fig functions that access dictionaries as well.

Years ago I created a simple hack with inline Python to access the 16-colour palette (fig uses foreground/background 0-15/0-15, based on the CGA palette) and allow fig to do 24-bit colour without adding commands. I decided for many reasons I preferred this hack to making it a feature of fig -- but I made it a native feature in figplus:

function rgbcolor r g b
python
    figcgapal[0] = (r, g, b)
    fig
    fig


This changes colour 0, so the next time you draw it will use whatever RGB settings you like. You can change it back to 0, 0, 0 after that if you want.

I made four changes to the stable version of fig for the latest, and each change is related to GitHub:

1. Pygame is gone, even as an optional feature. This is a protest, if someone wants to fork fig to keep Pygame I'd be happy to help them do it. It's simple enough that they probably won't need it. The change is backward compatible -- commands that only did something when Pygame was installed, now do nothing even when it is. No code changes should be needed.

2. PyPy 2 is now used instead of CPython. This was likely to happen anyway, because the Python Foundation wants to force a language I don't care about. I knew a guy that worked at Nokia who felt the same way, but there are plenty of sound technical treatises on why strings are broken in Python 3. Don't agree? Then use what the Python Foundation tries to shovel at you -- most people do.

"...not only am I not interested in Python 3, it is also developed on Microsoft GitHub."However the reason is two-fold; not only am I not interested in Python 3, it is also developed on Microsoft GitHub. PyPy is not, and I hope it never will be. People who have Python 2 installed on the GNU operating system that want to continue to use it with fig only need to make a symlink; I'm not judging. If you have trouble with a symlink, copying your Python executable to something called pypy2 may help.

3. The arropen command already worked in both CPython and PyPy. However, PyPy doesn't close files read with .read() as I think it ought to, which means that after reading about 1000 or so files with arropen, PyPy complains about too many open files. It shouldn't be necessary, but fig 4.9 caters to PyPy by opening and closing files even when arropen uses .read(). As I said, the command works in both implementations. What's changed is that arropen should now work more reliably when under heavy use in PyPy.

4. I didn't just remove Pygame support, I tried to add something to make the GitHub boycott a little more fun. The hack that allows 24-bit colour in Pygame is now supported in text mode. If you don't include the hack in your program, figcgapal doesn't get modified (it can only be modified with inline Python; fig doesn't support identifiers that start with "fig") and fig does escape output the way it did before, with esc[0 or 1;(30 or 40 plus f)m.

"The screencap is also a jpeg, even though png would be better for this purpose. The reason is symbolic; png relies on zlib1g, which is on GitHub."If you use the hack, it changes the contents of figcgapal and fig uses esc[38;2;r;g;bm instead. This is designed to be compatible with existing code that uses the hack; fig now checks figcgapal against a copy that's made when the original is set. This was done in haste, and 4.8 is GitHub-free but the 24-bit feature didn't work until it was fixed (along with arropen) in 4.9, maybe 30 minutes later.

The screencap has fig 4.9 doing 24-bit output based on unmodified code, but I haven't tried it with a lot of code yet. The screencap is also a jpeg, even though png would be better for this purpose. The reason is symbolic; png relies on zlib1g, which is on GitHub.

Very few people have any idea just how much everybody is relying on GitHub right now. Although it may not be possible to get away from it (or Microsoft) completely, and removing Pygame from my favourite programming language may seem ridiculous, I don't think it's nearly as ridiculous as putting all our free software eggs in Microsoft's giant basket. We need to Delete GitHub. If it turns out that we can't do it entirely, we should at least consider our options. I'd rather sacrifice Pygame than use a canvas that Microsoft controls.

"Very few people have any idea just how much everybody is relying on GitHub right now. Although it may not be possible to get away from it (or Microsoft) completely, and removing Pygame from my favourite programming language may seem ridiculous, I don't think it's nearly as ridiculous as putting all our free software eggs in Microsoft's giant basket."To the Pygame devs: I love SDL (I love DOSBox! Which probably still uses SDL, but did when I created fig). I never preferred to use anything other than Pygame for fig graphics. I don't love the idea of trying maybe Tk instead. But Tk is at least responsible enough not to keep selling themselves to an aggressive monopoly that hates software freedom. I figure any graphics will likely pull in libffi (GitHub) though you're developing there directly. Please reconsider your reasons for this! Don't sell out your users to the GIAFAM trap.

Long Live Stallman, and happy hacking.

Licence: Creative Commons CC0 1.0 (public domain)

Recent Techrights' Posts

Rust People: Drain the Swap, You're Holding It Wrong
Does Rust make sense?
Slopwatch: LinuxSecurity, linuxconfig.org, and Plagiarised Phoronix
Many articles out there are nowadays fake
European Patent Office Illegally Gutting and Outsourcing Its Functions, Acting Like an Above-the-Law Commercial Business (It Won't Stop at Formalities Officers (FOs) and Classification Slop at the EPO)
breaking/violating laws and conventions
Links 19/09/2025: Lobbyist of American GAFAM Becomes Data Protection Commissioner in Europe
Links for the day
The Right to Punch People (Apparently)
At Brett Wilson, Brett's job title is "Head of Crime" and Wilson normalises calls for violence
 
About 700 New Gemini Capsules in 13 Months (or 54 Per Month)
4.8K would represent a 20% increase
Techrights the Name Turns 15
About 6 weeks from now we turn 19
Microsoft is Running Out of Time and Floating Fake Figures, Fake Projects, Fake Narratives, Fake Excuses
Also, a lot of Microsoft's "revenue" claims are circular financing (i.e. Microsoft buying from itself, which means Ponzi-like fraud)
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Friday, September 19, 2025
IRC logs for Friday, September 19, 2025
Gemini Links 20/09/2025: Navigating the Pressures of Modern Life and SpellBinding Accidentally Wrote Another Gemini Server
Links for the day
Links 19/09/2025: Press Freedom Dying in US, Anti-Austerity Strikes in France, and Alan Rusbridger to Leave 'Prospect'
Links for the day
Offloading to the Sister Site
In the interest of not overwhelming readers
Links 19/09/2025: Coffee Club and "SpellBinding is Now Absurdly Fast"
Links for the day
Links 19/09/2025: Media Freedom Ceases to Exist in US, "Consider Dropping Twitter/X"
Links for the day
Gemini Links 19/09/2025: Thinking and Insect Bites
Links for the day
Microsoft E.E.E.: Git Will Now (or Very Soon) Fully Depend on Rust, Which is Controlled by Microsoft
Microsoft now makes Git dependent on Rust, or making Git dependent on GitHub, which is proprietary
Slop or Fake Articles Have Turned Linux Journal From a Pioneering/Trailblazing "Linux" Magazine Into a Nuisance
some sites with former reputation - good reputation - turn into cesspools
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Thursday, September 18, 2025
IRC logs for Thursday, September 18, 2025
Brett Wilson LLP Seem to Have Had Only One Litigation Client in 2025, He Was Previously Charged, Just Like the Serial Strangler From Microsoft (Whom They Now Represent)
Karma is superstition, regulators are not
Project 2030 to Cover How "Project 2025"-Styled Anti-Media Zealots From America Targeted Techrights and Tux Machines
The common denominator is also their attacks on women
Brett Wilson LLP Failed to Meet Deadlines Set by Judge 7 Months Earlier, Tried to Ruin Our Holiday, Then Had the Audacity to Ask Us for Over 3,000 Pounds for Its Own Lateness
As a matter of principle we will never respond to assassin while we are on holiday
On Claims That After Bluewashing Red Hat Will Increasingly Become an Indian Company
Discussed this week (long and detailed)
Americans Attacking British Sites Only Months After They Leave America
We find it kind of funny if not ironic that this site, originally an American site, got legal harassment only from Americans and only months after it had moved to the UK
Despite Losing Over a Quarter Million Dollars a Year Software in the Public Interest (SPI) Gives Helping Hand to Libreboot
SPI's financial state depends a lot on its public image or its reputation
Slopwatch: Google Helps Plagiarism and Sends Traffic to Ripoff Artists
That Google as a company helps spamfarms is noteworthy
If You Want to Know the Future, Listen to the Free Software Foundation (FSF) and Andy Farnell
We're sure the FSF will have plenty of its own output
Links 18/09/2025: A Taliban Ban on Internet Access and Troubled US Job Market
Links for the day
Gemini Links 18/09/2025: Computer Literacy and Accessing Alhena's Database
Links for the day
Links 18/09/2025: US War on Media (Truth Banned, Cancel Culture by the Hard Right), NYT Chief Executive Warns Cheeto is Deploying ‘Anti-press Playbook'
Links for the day
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Wednesday, September 17, 2025
IRC logs for Wednesday, September 17, 2025