[openspending-dev] Slimming down the frontend

Mon Dec 29 17:41:51 UTC 2014

Hey all,

while refactoring the OS front-end to Flask, I've had a chance to refresh
my memory regarding all the bits of OpenSpending which might be considered
"there for legacy purposes". I just wanted to share some notes on things
which I think we should remove in a breaking fashion. Some of them would
have replacements, others wouldn't.

## Remove without replacement

* Old 'views' system, i.e. the JSON-configured views. This is nearly
incomprehensible, introduces nasty forms of state in lots of pages and
yields illogical visualizations. I'd like to propose removing it w/o a
replacement.

* /api/1/ - this is fait accompli, I didn't port it to Flask. It's been
deprecated for three or four years, time to go.

## Replace with a newer version

* There'll be lots of changes to the dataset editor and dataset index based
on the outcome of our discussion about metadata, master data, etc. etc.

* Move all the dependencies for page functionality in openspendingjs to the
main repo. I know we want to modularize, but this is the wrong way to do
it. In particular, I'd love to talk about ways in which we might be able to
have a more coherent angular app for the front-end, rather than 100 small
snippets of jquery.

* Replace treemap implementation with d3-based version. I already have an
implementation of this. Downsides: kills IE8 support, upsides: half as
large, fewer dependencies for the platform as a whole, looks nicer.

## Things I'm on the fence about

* Global search -- it's kind of cool, but if we remove it, we can switch to
database-based fts, from solr. Life would become so much better... If
someone then wants to make a spend-search service, no way to stop them :)

This is just a heads-up, might create tickets if there is no riots?

Cheers,

- Friedrich
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/openspending-dev/attachments/20141229/0f1cc758/attachment.html>