Templating engine woes about psij-python

[hategan]

I wonder though if we might consider using jinja instead?

We want a cross-language solution.

[hategan]

I can fork an updated version, tag, and we can use that until we either move to a different implementation, different engine, or the one currently on pypi gets fixed. Does that sound reasonable?

[jameshcorbett]

If you want to take on the responsibility yourself, whatever you want to do to get things working is fine by me. I just want things to work, and I would love to get the readthedocs build to pass so our CI will actually turn green.

Our mustache options don't seem great to me at first glance, although they might turn out well upon testing. We could always just try out chevron or the new pystache and see how it goes, and whether the developers are responsive, and then decide if we want to use one of them.

I'm not convinced that we should really care that much about a cross-language solution. We're not using our templating libraries for anything very complicated so it's generally a pretty straightforward translation from one templating library to another (are the delimiters {% or {{ or %%?). My guess is that re-writing the templating stuff would be like 5% of the work of a new PSI/J language implementation.

But mustache is what we have now so we should try to make it work.

[hategan]

I'm not convinced that we should really care that much about a cross-language solution. We're not using our templating libraries for anything very complicated so it's generally a pretty straightforward translation from one templating library to another

It's an unnecessary step for users on a system that needs customized templates to have to do so for every language implementation and test such copy separately. We may never really get to that point, but, if we do, I'd rather not be in that situation because of a temporary snag with a library. It may be useful to recognize that we cannot, in general, guarantee that any of our dependencies will always be nice and dandy.

[hategan]

A recent version of pystache is now available and PR #100 switches to it.

[hategan]

#100 was merged into main, so this should be fixed now.