Need node IDs to be able to create more than one node with the same label

[lorlik]

Nomnoml syntax and graph rendering relies on node label for identity. Unfortunately, this is very limiting as there are many situations where one may have more than one node with the same label. Would it be possible to extend syntax and rendering to allow for optional IDs for all types of nodes (e.g. [<style> {id} label ]) where if ID is missing then the graph would continue to use label as ID, if its present then ID would be used as identifier of unique node instead of label? Use of curly brackets for syntax is just a suggestion, any syntax to indicate ID would be fine.

[skanaar]

I am right now doing a big rewrite of the parser to support some new features, for example an ID field.
But it might take a while, I have very little time for my hobbies right now.

[lorlik]

What was your vision for ID syntax in case we decide to contribute and add it ourselves?

[skanaar]

You can track the rewrite from Jison to Peggy in the rewrite-grammar branch.
https://github.com/skanaar/nomnoml/blob/rewrite-grammar/src/grammar.pegjs

This is the new syntax: [<type id=foo>foobar].

The plan is to allow any metadata to be added as key-value pairs within the angle brackets.

[skanaar]

Have a look at https://stage.nomnoml.com/ where I have deployed a release candidate for the new parser.

I would be very grateful for some testing here before I deploy it to www.nomnoml.com 🙏

[lorlik]

Thank you. I am getting a compile error on stage for a graph that works in the current version. There are no details to where the issue is. How would you like me to communicate the doc to you for troubleshooting? I'd prefer for it not to be in the public domain.

[skanaar]

You can email me on [email protected]. I'll delete it once I have identified the troublesome syntax.

[tcbuzor]

I tested the new parser for my current use case and it works. Do you have any other cases you need help testing?

[skanaar]

@tcbuzor nice to hear that it works.

Just test it with whatever diagrams you usually write. You might be using a different "style" than me, one that I haven't anticipated. 😄

[tcbuzor]

Sure, will test other scenarios. Do you mind also providing some insight into how one might implement the ERD associations? #185

[bdlabs]

Maybe this syntax would be a better way
#attr foobar: id=foo;
or
#.foobar: id=foo; in css

[foobar]

Because this solution will be clearer when the engine will be expanded with additional arguments like this.

[wurst-hans]

You can track the rewrite from Jison to Peggy in the rewrite-grammar branch. https://github.com/skanaar/nomnoml/blob/rewrite-grammar/src/grammar.pegjs

This is the new syntax: [<type id=foo>foobar].

The plan is to allow any metadata to be added as key-value pairs within the angle brackets.

Your project is really fantastic. I'm just looking for a simple JS library that is able to render a simple static flowchart out of a text definition. There are other really great projects like MermaidJS or jsPlump, but they are so huge and massive overkill. And your new parser was the feature I'm waiting for. Using numeric IDs for connecting elements, ex.:

[<note id=1>First slide]
[<start>Start]->[1]

is much easier than using long text labels. I love nomnoml.

EDIT: The only thing I had to change is exporting layouters because I need a minimum width and height of note boxes, so I've overwritten nomnoml.layouters.note as workaround.

[skanaar]

New version deployed to https://stage.nomnoml.com
It fixes the issues reported by @lorlik

And thank you @wurst-hans for your kind words 😍

[lorlik]

@skanaar , thank you for fixing the issue.

[42sol-eu]

Have a look at https://stage.nomnoml.com/ where I have deployed a release candidate for the new parser.

I would be very grateful for some testing here before I deploy it to www.nomnoml.com 🙏

@skanaar I will do that. My nomnoml skills have been called magic over the last years, this is because my brain really loves and integrates that syntax so much more than plantuml or mermaid.

and yes I have checked this with mermaid some weeks ago
my logical part of my brain says: "Mermaid really makes sence"
but my system 1 (see the book "thinking fast and slow") does not put
so I am still a believer of nomnoml that I can use whilst talking and in meetings as a magical white board.

If there are specifics to test please feel free to ping me!

[42sol-eu]

...
is much easier than using long text labels. I love nomnoml.

There is a trick that I found with the old parser:

1. Pick a short title for the first line (like [N1;this is the note])
2. now N1 is the id! use it like [N1] -- [cool]

[zflat]

You can track the rewrite from Jison to Peggy in the rewrite-grammar branch. https://github.com/skanaar/nomnoml/blob/rewrite-grammar/src/grammar.pegjs

This is the new syntax: [<type id=foo>foobar].

The plan is to allow any metadata to be added as key-value pairs within the angle brackets.

It looks like there are restrictions on what can be added as values?

I tried [<class id=x0_y0>A] on https://stage.nomnoml.com/ and get a compile error:

Expected ">" or space but "_" found.

Right now the value is restricted to characters [a-zA-Z0-9] but maybe some other characters could be added? In the new grammar, an Attr would need to be key:Ident "=" value:AttrVal where AttrVal would be a new definition that allows more charactes than the Ident does. Does that seem possible?

Edit:
Looking more at the grammar I realize that the attr value needs to able to be used as an Ident, so maybe Ident could also include _? I just think that it will be hard to write identifiers that can be readable only using camel case.

[skanaar]

@zflat good feedback.
I would like to keep some restrictions to which characters are allowed to allow for new syntax in the future.
But [a-zA-Z0-9_] seems reasonable. :)

[zflat]

Thanks! The only other thing I would request that could truly allow any metadata would be to support enough characters for base32 or base64 encoding, however supporting the = for padding probably makes this infeasible. I acknowledge that this is scope creep and most likely something you don't want to support so either way, thanks for all the work you put into this library and tool!

[skanaar]

I have deployed a new version to https://stage.nomnoml.com that solves some performance issues.

I would really appreciate if you can guys can test it out. As soon as I am confident it works it'll be deployed to nomnoml.com

[zflat]

Working for for my use-case:

Thanks again!

[skanaar]

New version released 🎉