Re: [xml-dev] RE: Schemaless XML?

["Simon St.Laurent" <simonstl@simonstl.com>]

Finally, an xml-dev thread that brings me joy!  (Not sarcasm. Real joy.)

On 10/12/2016 8:36 AM, Ihe Onwuka wrote:

No and this is what I tried to convey in my original (perhaps too pithy) response. Schemaless means  you can load your data into the repository without defining a schema a priori (in contrast to relational).  

That is as earth shattering and profound as being able to  put data into Excel without defining the column headings or having any at all. 

That's an excellent comparison.  While I spend a lot of time in Excel spreadsheets that are tightly structured - and frequently wish at least some of them lived in even more structured relational databases - there are endless cases where being able to just doodle in the spreadsheet is what gives a spreadsheet power.

Schemaless really means schema deferred. It gives you the luxury of allowing schema definition to leak into your application code so when someone asks you for the business rules governing your data you give them your (these days _javascript_) code.  

"Deferred" is an excellent way to put it. It's not so much that no one will ever want to apply a schema to the data or derive a schema from it.  Rather, it's that decisions about schemas can come later, in the field, made by people consuming or creating documents.  They may not even be formal decisions.

On single-person projects, that deferral creates a lot of freedom without much mess.  Even in larger projects, component approaches can create more flexible spaces within larger structures.

I think we may finally be reaching the cultural point in computing where those freedoms are appreciated rather than cursed as disorderly.

Thank you all,
Simon