Hi Bento,
for the programming background of your idea, I humbly suggest you have a
look at Metafont, Donald Knuth's font description language created by
him for the TeX system, as a source of conceptual inspiration.
Besides the more conventional way of assigning values to variables,
Metafont offers a second mechanism which may be helpful for your project
as it contains an equation solver. See page 14 of this tutorial:
http://metafont.tutorial.free.fr/downloads/mftut.pdf
Your ideas outlined below immediately prompted me to think of Metafont.
The linear equation solver is also part of the MetaPost programming
language; Metafont is open source, the source file is well documented.
The code itself is written in a well-readable meta-language.
Unfortunately I stopped programming in Metafont nearly two decades ago
and am currently engulfed in a gyre of work, so I cannot offer more
immediate support; I only can wish you
"Good luck on your mission!"
Best,
Oliver.
On 29/11/2025 05:10, Bento Borges Schirmer wrote:
Hey list,
I really really really really liked pic.
This vacation I will create a small tool that reads a fraction of pic
syntax, and turns it into C code, that can be compiled to draw GUIs in
a graphical program.
My idea is to have a fully imperative GUI builder.
So, for example. A windowed app, the window is resizable. As it
shrinks, elements cease to fit inside, so you need to replace them
with a smaller version, or collapse them entirely inside a menu. I
want this to happen fully imperatively, like, there is a condition
that tests if things fit or not, and what will be drawn next. Another
case would be layouting a grid, I want the layout to be decided
imperatively using a linear or binary search. All of this exposed to
the programmer, no abstraction.
I want it to contrast with current declarative means to create GUIs.
You see, I have zero experience working as mobile or web developer, so
I don't have experience to back up any claims; I can't say from
experience that such and such is bad or good. But I get the feeling
that the abstraction provided, you have to fight with it a lot, like,
translate what you really want "position this box relative to this
box" with the hierarchy of elements it imposes, and whatever automatic
means it has.
The few times I tried to build GUIs, I wanted to position stuff in
certain ways, like in the bottom of the window, and I'd be told "just
calculate everything yourself then". That sucks, because the solution
is not really helping at all. pic at least have several means to
specify things relative to each other, and group things arbitrarily,
so there is that.
But the idea is to extend the pic syntax with clauses to "give up" a
block of graphics, and a "eval" that will just calculate bounds
without drawing, and some other that "branches" several attempts and
only draws some, but also memorizes other results somehow, and there
is also scrolling to deal with.
The point is. I will implement a basic fraction of pic, if it works, I
will try to expand more. Or maybe I misunderstood everything! I also
don't have experience with GUI development, so I may also
misunderstand the "real" challenges current technology really solves,
like uuuuh... I don't know?
I've been using groff heavily to produce my thesis text and slides,
after my semester ends, I have plenty to discuss, lots in my mind
right now.
Best regards,
Bento
--
Dr. Oliver Corff
mailto:[email protected]
