Ubuntu moves to GTK4. Forum MUST consider

[Clarity]

For new WoofCE PUPs based on Debian/Ubuntu, it must consider when it should move too.

Arch has already done this many months ago.

Even if some disagree, it is important to consider forum timings to move on from GTK2+

FYI

[wiak]

For new WoofCE PUPs based on Debian/Ubuntu, it must consider when it should move too.

Arch has already done this many months ago.

Even if some disagree, it is important to consider forum timings to move on from GTK2+

FYI

I think you'll find that is more to do with dependencies of apps installed. Certainly, some apps might only be compiled by Ubuntu with earlier versions of GTK, and a distro builder could then select not to install some apps to avoid having more than one GTK version, but overall, if distro repo provides apps dependent on GTK4 then GTK4 is going to be installed or will need to be if package manager itself does not pull in required dependencies. Certainly if trying to create a GTK4-only distro then the likes of Puppy's version of gtkdialog would need its source code checked/altered such that it compiles successfully with GTK4 and worked without issues with Puppy utility apps that need it and so on.

Puppy doesn't usually come with a GNOME desktop so the fact Ubuntu likely to use GTK4 version of that in releases after Jammy doesn't really matter. Its use still doesn't mean all apps will necessarily be compiled to use GTK4 anyway.

[dimkr]

The migration to GTK+ 4 is more than installing a package. It won't just happen.

gtkdialog needs to be ported (and many things will break, some widgets are gone), themes need to be ported, all applications traditionally included in Puppy (like Geany) need to be ported and so on.

Basically, it's like the GTK+ 3 migration (of dpup and jammy64) all over again, but much harder.

[wiak]

The migration to GTK+ 4 is more than installing a package. It won't just happen.

gtkdialog needs to be ported (and many things will break, some widgets are gone), themes need to be ported, all applications traditionally included in Puppy (like Geany) need to be ported and so on.

Basically, it's like the GTK+ 3 migration (of dpup and jammy64) all over again, but much harder.

Indeed, lots of porting involved, so maybe Puppy Ubuntu variants will stick with Jammy for a long time (as they have done with FocalFossa). Thereafter will depend on who is willing to get stuck into woof-CE (and gtkdialog repo and so on) and do the actual porting. Whether it will happen depends on who takes it on and their abilities. It is not particularly easy upgrading Puppy methinks, but always possible, albeit not without hands willing to do it. Still a fair number of Puppy users, some been around here for a long time, but more than user-enthusiasm and requests will be required - contributors able to do the work are always needed to prolong any and all distros lifetime. Okay maybe for next few years, but five years from now it depends on overall contributor activity. I'm not the nostalgic type so feel it doesn't really matter if all of Puppy, KL, FirstRib, FatDog, DebianDog or whatever reached the end of line in terms of their active development - there will always be some version of Linux - it is the overall technology and its usefulness that is important to me not any particular distro. One thing I'm pretty sure about is that five years from now I do not imagine myself to be doing anything much active in any kind of computer development work, but who knows... time flies and five years away will soon be here. Maybe AI will churn out all the distros we might want by that time... so humans can get on with work more social and important to us as individuals???!

[geo_c]

Maybe AI will churn out all the distros we might want by that time... so humans can get on with work more social and important to us as individuals???!

That's a really deep reality that gives me pause.

Think about all of the social interaction, (even though online with people we've never met in person) that occurs when humans team up to develop and tinker with their tools.

Perhaps @Clarity will learn how to coax puppies, first-ribs, and other tools from GPT, and then none of the discussions, disagreements, and social interaction found in online forums will be necessary. We will simply type a prompt: "ChatBot, compile a puppy linux using Ubuntu gtk4 with the following utilities and applications."

And at that stage of society we will never know that people like @wiak and @dimkr ever existed in places around the world we've never visited, who have unofficial opinions and knowledge that ChatBot will never impart to us.

Upside I suppose is people will be able to spend more time with flesh and blood relationships. But will they? Some are already involved in romantic relationships with their ChatBot buddy.

Not to derail the Gtk4 issue. It seems to me that first-rib has an advantage in that the script builds the system "up-to-date," though many or most beloved puppy utilities may not be able to be added, and in that sense is not "puppy" as @wiak is sure to agree, but still has many of the installation/portability features of a puppy, and that perhaps might be the endgame if more and younger puppy developers aren't forthcoming.

Since I have absolutely no coding/scripting training and can barely compile a package, only having learned how to do it this year, I know it's not going to be me building attractive and versatile distros using woof, though perhaps it would be possible for me to run a first-rib script successfully.

Try asking ChatBot: "Build me an Ubuntu Gtk4 disturbution packaged as an iso using First-Rib available at the following url:"

[mistfire]

Regards to porting, what if using ChatGPT in order to port Puppy GTK+2 apps into GTK+4 apps faster?

I successfully forked and ported gnome-schedule from python2-pygtk to python3-pygobject with the help of ChatGPT as assistance

The trick here was ask ChatGPT to translate a source code from GTK+2 to GTK+4 equivalent. Some lines of code at a time (not the whole source code) and make some tweaks.

[dimkr]

AI-assisted programming has its limits. It's good for trivial codebases where you can see with your own, unintelligent eyes, if the generated code makes any sense. Tools like GitHub Copilot sometimes copy badly written code and can copy a bug or a security issue. And they don't understand software licenses, copyright, readability or the idea of well-structured or compact code.

AI probably won't help much with refactoring of GTK+ 2/3 applications that need UI redesign (menus -> popovers, etc') and UX work.

[dancytron]

The migration to GTK+ 4 is more than installing a package. It won't just happen.

gtkdialog needs to be ported (and many things will break, some widgets are gone), themes need to be ported, all applications traditionally included in Puppy (like Geany) need to be ported and so on.

Basically, it's like the GTK+ 3 migration (of dpup and jammy64) all over again, but much harder.

Indeed, lots of porting involved, ***
. I'm not the nostalgic type so feel it doesn't really matter if all of Puppy, KL, FirstRib, FatDog, DebianDog or whatever reached the end of line in terms of their active development - there will always be some version of Linux - it is the overall technology and its usefulness that is important to me not any particular distro.

***

I'm not particularly nostalgic, but layered file systems with read only files and controlling when things get saved from ram to disk is just the correct way to set up a small computer. It just is. I think Debian Dog and KLV have shown that is going to be viable for the foreseeable future without relying on custom kernels and aufs, which is reassuring.

[mistfire]

AI-assisted programming has its limits. It's good for trivial codebases where you can see with your own, unintelligent eyes, if the generated code makes any sense. Tools like GitHub Copilot sometimes copy badly written code and can copy a bug or a security issue. And they don't understand software licenses, copyright, readability or the idea of well-structured or compact code.

AI probably won't help much with refactoring of GTK+ 2/3 applications that need UI redesign (menus -> popovers, etc') and UX work.

That's why I'm telling to ask an AI to translate only a parts of source code not a whole source code paired with reviewing the code generated by AI because of that limitations

[pp4mnklinux]

May 07

viewtopic.php?t=256&start=940#p88427

3x

AI-assisted programming has its limits. It's good for trivial codebases where you can see with your own, unintelligent eyes, if the generated code makes any sense. Tools like GitHub Copilot sometimes copy badly written code and can copy a bug or a security issue. And they don't understand software licenses, copyright, readability or the idea of well-structured or compact code.

AI probably won't help much with refactoring of GTK+ 2/3 applications that need UI redesign (menus -> popovers, etc') and UX work.

[Clarity]

Hi @pp4mnklinux

I think @mistfire is showing is that a develop who masters ANY tools, also will master methods of development-repair when necessary. All good developers do these things. There will be many methods to debug code/logic similar to the ways we do things today.

The corporate world, which sponsored AI advancements will create methods that will address how to debug-repair-redesign every creation the corporate produces. Their development-engineering teams are good at it.

The post does not have anything to do with this topic as it is only serving as "to consider timings".

One thread on this forum is discussing "AI technology" where members are offering AI ideas and findings. That forum thread is here.

[pp4mnklinux]

It is possible it was my error, because I understood the pic other way... so excuseme not to see this photo only serving as "to consider timings".

Hi @pp4mnklinux

...

The post does not have anything to do with this topic as it is only serving as "to consider timings".

Have a nice day, @Clarity

[dimkr]

Hi @pp4mnklinux

I think @mistfire is showing is that a develop who masters ANY tools, also will master methods of development-repair when necessary. All good developers do these things. There will be many methods to debug code/logic similar to the ways we do things today.

The corporate world, which sponsored AI advancements will create methods that will address how to debug-repair-redesign every creation the corporate produces. Their development-engineering teams are good at it.

The post does not have anything to do with this topic as it is only serving as "to consider timings".

One thread on this forum is discussing "AI technology" where members are offering AI ideas and findings. That forum thread is here.

This reply looks like AI-generated content. I'm a programmer for living, I have no idea what "debug-repair-redesign" means, and I never heard this term in job interviews.

[Clarity]

Hi @dimkr Hoping you got the point. We dont speak the same languages. You may not have heard it that way, but if we can get past that language barrier, I think much progress can be made.

[dimkr]

Please define "debug-repair-redesign" so we can understand each other.

[amethyst]

Please define "debug-repair-redesign" so we can understand each other.

Probably debug then repair then redesign is what is meant. debug-repair-redesign is just a sort of Clarityism (which MUST be considered).

[wiak]

So, what seems to me a reasonable interpretation:

A system develops a fault. The engineer thus has to debug what is causing the fault and thus find a suitable repair. After doing that, the good engineer/designer/developer will likely think, hmmm, I don't want that fault to arise again - lets redesign that part of the system. Result (all going well): an improved design and a more fault-tolerant system.

[sonny]

AI-assisted programming has its limits. It's good for trivial codebases where you can see with your own, unintelligent eyes, if the generated code makes any sense. Tools like GitHub Copilot sometimes copy badly written code and can copy a bug or a security issue. And they don't understand software licenses, copyright, readability or the idea of well-structured or compact code.

AI probably won't help much with refactoring of GTK+ 2/3 applications that need UI redesign (menus -> popovers, etc') and UX work.

That's why I'm telling to ask an AI to translate only a parts of source code not a whole source code paired with reviewing the code generated by AI because of that limitations

Have you guys tried an "AI for developers" called CodeWhisperer?

https://aws.amazon.com/codewhisperer/