Soon as I can I am going to reply your posts.
THanks!
I do, but it is not about me.
This topic is about better ways to design and write software.
Especially about code bloat, excessive bug fixes and eternal updates.
I am still looking for something more radical. … I think we need a paradigm shift here.
When it reaches the stage that it is not feasable to compile it , except on a supercomputer, then I think we have to look for another direction.
The compile time issue is part of the reason for snaps and flatpaks .
There is a limit … the binary, with all its dynamic libraries loaded has to fit into ram. That used to limit programs on old mainframes. We have pushed that limit back with modern cheap ram, but the way we are going software will catch up. The day of the 64GB binary is approaching.
Even AI code will not solve that problem … it will make it worse. From what I have seen of vibe-coded projects, AI pays no attention to size of binaries or design of modular code.
I like the idea of Vim and Vim-Tiny.
Yes, why not have a tiny version as well as the full version.
That would work for any app.
I actually use vim-tiny or original vi, depending on the distro.
Yes there is a tiny version of X11 … used in tinycore linux and dsl.
Puppy Linux also reportedly has a tiny X . I didnt check.
When I ran my computer company in france I signed up to be a microsoft business partner as they were lacking my skills knowledge and certification. They very quickly agreed. But as time passed they started to askmore and more of me and my company. Eventually I dropped them as they wanted too much, so I went fully linux.
My last conference was in Bordeaux where they wanted me and other like me to develop applications to sell, I said NON !, but they were looking for android type apps on windows tiny things to sell and I would get the sells of the peanuts.
But like most of us I have a phone and tablet with apps on which take almost no space, fast and portable some of these should be available in linux to make it better.
Take word processing word is expensive, libreoffice is great for free but still mamoth in size. We need little tools that do the minimum and no more.
Agree. Browsers, DE, Liberoffice and a lot of other programs (app) are not only too large but have way too many options. I have to hunt and peck to find the right knob I want some times. Maybe an option is to have 2 versions of a program. One basic one and the other with all the bells and whistles that anyone would ever want.
To some extent we had that with microsoft office and microsoft works but works now has gone and I have never seen a linux type replacement
You are right , Android apps are tiny and easy to learn and use. What they lack is communication between apps, and that is a deficiency in Android.
Linux could use more tiny apps and they could potentially communicate better in Linux.
The vim and vim-tiny idea solves that, but there is also a case for better menu design. … it is possible to have 2 levels of menu … basic and advanced … but few apps do that.
If I were writing Gimp, I would make every image processing item in the menu a separate Linux command and use the Linux system to pass images between commands. Then Gimp would be like a language instead of a monster collection of menu items.
Languages are more flexible than menus. You can have a help function to find commands. With menus you have to hover around to find commands.
How could we make the actual browser smaller without it losing functionality? HTML, CSS, JavaScript, and all the other functionalities (and there are a lot of them) are tightly integrated. I don’t see a way to break that up into smaller pieces without affecting functionality or performance.
Same goes for word processors. LyX, for example, is quite small, but it uses LaTeX as a back-end and is non-trivial to use.
Thing is, some way needs to be found to split all these big programs up, but still keep them useful. So, it becomes the question how to go about this without damaging functionality and performance?
This is the key. I don’t need or want all of this functionality. But I’m a victim of the one-size-fits-all paradigm.
From my working life: Every software developer knows how hard it might be to say no. Hey, maybe the proposed feature could be cool and/or helpful in getting better than the competitors. Hey, please add this… Why can it not be used for making Greek-style yogurt? Let’s go!
For the text based programs there’s the stdout, stdin, stderr as channels to send processed informations, status info, and so on.
For the GUI I think dbus is (could be?) usable.
I think so,but it is system level stuff … it needs developing, plus the apps themselves need a reorientation
Exactly.
We have all become addicted to these ‘do everything’ apps, because they do really make activities like browsing easy.
So that success makes it hard for us to conceive of another way of doing things.
There will be a breakthrough . What a browser does is not all that complex. Someone will find a smart way to shortcut it.
You may have answered @xahodo here.
I need a browser targeted to my needs and personality … so do you, but yours will be different.
So how do we get a universal browser base, plus user-specific addons?
Would that base still be too large and bug-ridden?
AFAIK, dbus is that sort of pipe, but the communication overhead of such mechanisms is enough to slow things down and responsiveness is a highly valued browser metric.
Correct.
I think dbus is only intended for tiny communication tasks … not massive input or outout files.
It is not convenient either … requires programming .