[conspire] XML Considered Harmful (was: Acronym expansion, taking pity on the general reader)

[Deirdre Saoirse Moen]

On Fri, Sep 25, 2020, at 2:09 PM, Ivan Sergio Borgonovo wrote:
> I'm a bit surprised. I've been a drupal user and developer and I got fed 
> up with the sect of some core developer and by the never settling API.
> I grew up with the assumption that a large part of WP success was due to 
> a smooth path of upgrade (core and modules).
> In fact I think Drupal lost a big slice of market to WP exactly for this 
> reason.

I agree with you, but part of that was also that, because PHP was always ugly and stayed that way, it had a smoother upgrade path than the incredible speed bumps in Ruby/Rails (as an example).

> What's sad is there seems to be no alternative in that software space.

While there are other tools, I decided to go with Jekyll in part because that's what GitHub uses for GitHub Pages, and is thus worth learning just for that reason.

1. Markdown (easy peasy)
2. Static generation of sites

But it has all the Ruby annoyances too, including brain damage about 1.2.7 not being greater than 1.0.0. :P

> As a side note:
> 
> xml: it offers flexibility where no one is willing to use it. It has 
> Xpath that can be handy when your software/module is 
> responsible/interested in just a section of a file. I think I played 
> with xls transformation more than 20 years ago but they seems too 
> complicated. I'm not sure if anyone is still using xml/xls this way 
> unless for legacy reasons. Again Xpath can be handy and I'm not aware of 
> any equivalent.

I used to do a lot of XSLT and so I'll tell you the big problem with XSLT/XPath: the order of operation is not inherently defined. You have two things that could match a path, you just don't know which is going to happen first, because that's not in the spec. (Assuming the path elements are at the same level in the hierarchy.)

So it's a blessing AND a curse, and it leads to some remarkably subtle bugs that probably won't be found for years.

You *can* sort, it just can muck with other things you can do because WTAF. 15 years ago I did several web apps with XSLT transforms and I'm breaking out in hives just writing this. :P (EOModeler, Java, on a then-current MacOS X)

Despite my issues with XSLT/XPath, I actually really liked that you could take results and 

> json is mostly OK. Not too different from many way to serialize data 
> across many different languages but very well supported by most 
> languages. Native to JavaScript. JS should die together with Flash but 
> that's what we have.

JSON is a vast improvement, I will grant you.

> yaml: less typed than json but way more fun to read and write. Solve 
> most of the problems.

YAML version 1.2 is actually a superset of JSON. https://en.wikipedia.org/wiki/JSON#YAML

Deirdre