Introduce text in steps:

• fixed max length
• monospace
• text should fit in all types of steps (lower and upper case)

Time to extract some modules (e.g. constants, shapes, etc.). Cabal project seems reasonable.

Optional, but would be really good to learn it instead of depending on chained cli commands.

Now that I'm preparing to implement decisions, a tree would be a much better data structure to represent diagrams.

properties:

• name
• origin (Point V2 Double)

uniqueCoordinates :: Tree Step -> Double -> Double -> [(OriginCoordinates, Diagram B)] -> [(OriginCoordinates, Diagram B)]

Should ideally be converted to map as rendering can get expensive. Don't touch until we have:

• good telemetry
• significantly more complex diagrams

Replacing lists with maps should bring performance benefits, let's just make sure we can measure them first.

Some good examples here: https://drakonhub.com/drakon-examples

Buy a puppy example looks especially interesting - relatively simple and easy to render using just sdoc ppr.

source

At the moment I only have one parent per icon. Drakon diagrams allow for multiple connections to many parents, so this rendered will need it.

Let's start with one end icon being reachable from different steps.

All step types bar the start type can have multiple steps connecting to it.

The challenge is going to be to:

• position
• connect

them all.

What would be very handy is a debug mode in which we should render:

• icon origin (icon marking where the rendering originates)
• icon payload
• rendering order
• max width

Something like this: source: https://drakonhub.com/how-to-flowchart

• decision rendered - single outcome, treat as a regular command step
• chain: command -> decision -> command, fixed layouting
• chain: command -> decision -> decision * N -> command, fixed layouting
• automatic layouting
• step dimensions expressed as cell proportions (already done with command steps)
• automatic custom connections
• decision outcome connections to next steps should be decorated with yes/no

where N is an arbitrary, finite, positive number, say in range 0-10.

• introduce start and end steps
• connect them to regular steps
• rethink the idea of a regular step, we probably need something more specialised

before:

Today we have:

String -> [Icon]

as input. Now we need:

String -> [Icon]

so that we can move on to actually formulating functional graphs from the parsed strings.

Let's treat 10 as a magic number for the current stages of the project. Once more mature, the I can start experimenting with raising that number, but for now it's sufficient.

Now it serves no purpose, steps don’t need names in the current implementation.

At the moment rendered icon messages are displayed like this:

"quetion 1 | rendering order: 3 | max width: 0.0"

we can probably do better by displaying that information in multiple lines:

"quetion 1
rendering order: 3
max width: 0.0"

As a result, we could fit more text per icon or increase font size.

At the moment the connections need to be performed one by one. It would be good to automatically connect all steps.

At the moment there is only one graph and it’s all hardcoded. It would be very convenient to have some sort of input graph string to enable quicker iterations.

I feel I could use more guidance regarding code formatting and linting. It could also be useful for ci. I read here: https://analysis-tools.dev/tag/haskell that hlint is recommended and it also looks to be easy to install since we now have cabal fully leveraged.

https://github.com/ndmitchell/hlint/blob/master/README.md#installing-and-running-hlint