Version 4.0 ideas

[jquorning]

Ideas for Yass 4.0:

• Use markdown crate instead of libcmark. No more dependence on C.
• Allow spaces around variables. Easier to read.
• Rename variables to ex. site.description, page.contents, and page.title. Easier to read. Compatibility with Jekyll.
• Enclose variables in --- separation bars. Compatibility with Jekyll. Remove use of comments.
• Change format of site.cfg. Use : instead of = to make more like page style.

There is a problem with the new AWS crate Issue#2084. It will not install projects dependant on it.

Also there was a problem with gnattest, not allowing testing of procedures. There is a new AdaCore/gnattest repository on GitHub, but no Alire crate yet.

I have collected some Ada resources. It is built with Yass. jquorning.github.io.

[AJ-Ianozi]

I actually did make a branch locally on my computer of yass that uses the vss markdown crate instead. Unfortunately it's not fully compliant with cmark yet, which meant that websites previously generated with yass break. Trust me I've been wanting to get rid of that C dependency for years and have been considered just trying to write my own markdown library with the intent to be fully cmark compliant.

EDIT: I woke up and answered this, but didn't respond to the other stuff or expand on this. One of the biggest issues with the current markdown library is that it leverages regex heavily and can't just do "generic HTML passthrough" the way cmark can. Max has been very helpful and implemented support for many HTML tags, but it's not all the way there yet. It's a big reason I brought the libcmark in.

Can you show some examples of the variable names, spaces, and --- separation? I think it's reasonable especially if it doesn't change the way current stuff works too much (but I'm not against less extreme breaking changes since we are going to a new version in 4).

I'm still wondering if we should just have a standardized parser for the config file, either move to toml/yaml or at least decide on a standard like ini. Since cfg isn't really a standard, I've seen both : and = used, the latter usually leveraging INI format. Just have to think about people leveraging : in keyvalues such as https:// if we wanted to make that switch.

The biggest thing I'm working on for YASS4 is how it reads in the current site data. The current system just processes a page at a time without much knowledge of "the rest", but if we're going to allow proper plugin support (with HAL) and more extendable syntax, the new parser is going to have to "catalog" the whole site first, then process it. I'm still working that out.

P.S. I'm hoping the AWS Bug is fixed soon, there's a good chance it'll have to be fixed on the alire side with some kind of patch since AWS is EOL for the moment.