[Feature request] diagram options support

[yann-soubeyrand]

Hello,

As discussed in #333, currently diagram options (https://docs.kroki.io/kroki/setup/diagram-options/) aren’t supported. A use case this support would help with is to be able to specify the view of a Structurizr diagram to render (currently, the rendered view is random).

[ggrossetie]

The main question is what naming convention should we use?

• Use the diagram option names without suffix or prefix. In this case, how do we know which attributes are diagram options and which attributes are not? Would this cause any problems to send all attributes even those are not diagram options? We could use a pre-defined list but each time a new diagram option is available we will need to make a release.

  structurizr::partial$diagram.dsl[view-key=SystemLandscape]

• Use a prefix, such as diagram-option-?

  structurizr::partial$diagram.dsl[diagram-options-view-key=SystemLandscape]

• Other?

[yann-soubeyrand]

I personally don’t really like the first option, I prefer the second. Would it be possible to have something like

structurizr::partial$diagram.dsl[diagram-options="view-key=SystemLandscape,another-option=value"]

?

[ggrossetie]

Would it be possible to have something like (...)

That's another option. I feel like we are not using AsciiDoc with this syntax since we don't really use the attributes system. Some diagram options might expect a comma-separated list of flags/values. For instance, a font attribute where the value is the font name and the font size: font=Arial,11

We could workaround this "issue" by using an escape character \ or an another separator such as ;. Using an escape character might not work because it might be interpreted by Asciidoctor parser.

That's basically why I'm not a big fan of this solution because we are introducing our own syntax within an AsciiDoc attribute (and as a result, we will need a micro-parser).

I personally don’t really like the first option (...)

The first option aligns with Asciidoctor Diagram: https://docs.asciidoctor.org/diagram-extension/latest/#diagram-attributes May I ask why you don't like it?

(...) I prefer the second.

It could work but it's really verbose. Please note that we can't use options since it's already a named attribute in AsciiDoc: https://docs.asciidoctor.org/asciidoc/latest/attributes/options/

[yann-soubeyrand]

Would it be possible to have something like (...)

That's another option. I feel like we are not using AsciiDoc with this syntax since we don't really use the attributes system. Some diagram options might expect a comma-separated list of flags/values. For instance, a font attribute where the value is the font name and the font size: font=Arial,11

We could workaround this "issue" by using an escape character \ or an another separator such as ;. Using an escape character might not work because it might be interpreted by Asciidoctor parser.

That's basically why I'm not a big fan of this solution because we are introducing our own syntax within an AsciiDoc attribute (and as a result, we will need a micro-parser).

Yes, I agree with you.

I personally don’t really like the first option (...)

The first option aligns with Asciidoctor Diagram: https://docs.asciidoctor.org/diagram-extension/latest/#diagram-attributes May I ask why you don't like it?

I found the fact that we’ll had to either hardcode the list of possible arguments or to pass all the arguments including non diagram options not ideal. But if this is already done this way in Asciidoctor Diagram, maybe we should also go this way. Also, it seems that only the target and format atributes have to be filtered in their case. Would it be significantly difficult to filter out some attributes in our case?

(...) I prefer the second.

It could work but it's really verbose. Please note that we can't use options since it's already a named attribute in AsciiDoc: https://docs.asciidoctor.org/asciidoc/latest/attributes/options/

I may find the first option is the best in the end ;-)

[ggrossetie]

I found the fact that we’ll had to either hardcode the list of possible arguments or to pass all the arguments including non diagram options not ideal. But if this is already done this way in Asciidoctor Diagram, maybe we should also go this way. Also, it seems that only the target and format atributes have to be filtered in their case. Would it be significantly difficult to filter out some attributes in our case?

That's a good idea, it might indeed be easier to reverse the logic and filter built-in attributes 👍🏻

I may find the first option is the best in the end ;-)

Perfect 👌🏻

[sixtysecrun]

@ggrossetie I can see that you solved the problem of specifying view-key in #386 , but I don't understand how to use it 😞

Do you have an example of how I should specify view-key? I am trying the following ways 👇 , but none of it works. I am just served a random view.

[structurizr,diag1,svg,view-key=diag1]
----
include::diagram.dsl[]
----

structurizr::diagram.dsl[svg,view-key=diag2]

diagram.dsl

``` workspace { model { user = person "User" softwareSystem = softwareSystem "Software System" { webapp = container "Web Application" { user -> this "Uses!!!" } database = container "Database" { webapp -> this "Reads from and writes to" } } } views { systemContext softwareSystem "diag1" { include * autolayout lr } container softwareSystem "diag2" { include * autolayout lr } theme default } } ```

[ggrossetie]

@sixtysecrun Are you using the latest version of Asciidoctor Kroki? Please use the community chat to ask question: https://kroki.zulipchat.com/

[sixtysecrun]

Are you using the latest version of Asciidoctor Kroki?

I am using the latest version of Asciidoc Visual Code extension. I have just checked that it seems to work with Intellij Assidoc plugin.

Please use the community chat to ask question: https://kroki.zulipchat.com/

I will. Thank you!

[ggrossetie]

The Asciidoctor VS Code extension does not use the latest version that's probably why it's not working: https://github.com/asciidoctor/asciidoctor-vscode/blob/58adc632e6b9dfefd844efa66dd63ebdf9d1d78b/package.json#L658

Please open a new issue to request a version bump on asciidoctor-kroki.

[yann-soubeyrand]

Hey @ggrossetie, I realize I’ve never thanked you for this feature, so thanks a lot, it works like a charm!