search

schemeorg-community / monorepo

Scheme.org community subdomains that don't have their own repo
5 stars 4 forks source link

How to add auto-generated parts to web pages? #13

Open lassik opened 11 months ago

lassik commented 11 months ago

(Prompted by issue #11)

Many of our current pages are fixed .html files. It's not simple to add machine-generated information (e.g. standardi links and timestamps) onto those.

On the www.scheme.org front page, generated from the repo https://github.com/schemeorg/schemeorg, we use an approach where the HTML structure of the page is embedded as SXML in a Scheme script. Most of the text on the page is parsed from Markdown (.md) files into SXML and inserted into the right place in the structure. This works nicely. Could we scale up this approach?

arthurgleckler commented 11 months ago

I replied on #11, but should have replied here:

On another project, I've had luck using Pandoc to convert HTML into JSON, then reading the JSON, manipulating it, and converting back to HTML. We could use that technique to put footer elements even on static HTML pages.

lassik commented 11 months ago

Quoting @spdegabrielle:

I understand the urge to automate - but I believe it creates a barrier to contribution.

If I edit html I know it is going to work, if I use an automation someone else has put together - even common ones - you add the responsibility for me to locally implement your implementation to test my change has worked before posting the PR.

You also run the risk I am naughty and don't test my PR - then the repo owners get the additional work that comes with untested PR's failing more often than ones that have been tested.

lassik commented 11 months ago

You also run the risk I am naughty and don't test my PR - then the repo owners get the additional work that comes with untested PR's failing more often than ones that have been tested.

This happens already. It's not a big deal - one minute is wasted occasionally.

lassik commented 11 months ago

On another project, I've had luck using Pandoc to convert HTML into JSON, then reading the JSON, manipulating it, and converting back to HTML. We could use that technique to put footer elements even on static HTML pages.

Based on both your comments and my own recent experiences, I'm starting to lean toward this option.

So the repos would contain bare-bones HTML files that produce readable pages as-is, but are fleshed out with boilerplate and styling by script. @spdegabrielle Is this OK?