[LRUG] Config

[Jonathan Fantham]

I have come up with solutions to a couple of these which look a little different to what you’re doing:

2: I’ve often solved this by encoding config environment variables first into json, and then into base64. You get your booleans, hashes, or array of integers out like so: MultiJson.load(Base64.decode64(ENV[‘my_var’])). Json has always been enough for me to describe my config and you don’t have to do any manual parsing at all.

5 + 6: In the past, I’ve solved this problem by declaring a list of expected environment variables in an initializer (maybe you can get your list directly out of your config wrapper?). You can check if the variables are nil on startup and if they are you can raise an error. This will prevent your app from starting so you know immediately if there were any disconnects between defined environment variable names and expected ones.

I know that’s not an answer to your main question but you might find those ideas useful.


> On 27/02/2019, at 10:52, Patrick Gleeson <patrick.c.gleeson at gmail.com> wrote:
> 
> Hi LRUG,
> 
> After the awesome discussion on 'Continuous *' back in January, I wondered if this might be a place to ask another broad question I've been mulling for a while now.
> 
> According to the 12 factor app (https://12factor.net/config <https://12factor.net/config>) we should store our config in environment variables, and in Ruby that means we access it via ENV. I get the benefit of orthogonal config, but ENV always ends up feeling pretty unsatisfactory to me for config-heavy apps, for the following reasons:
> 
> (1) No easy way of seeing for any app what config needs to be set (cmd-shift-F 'ENV' anyone?)
> (2) Everything is stringly typed (How many times have I forgotten that 'false' is truthy? And what if a config var is going to be, say, either an array of integers or null? "&.split(',')&.map(:to_i)" is not my favourite thing.)
> (3) No single place to put sensible defaults if an environment variable not set
> (4) No namespacing or grouping except by prefixes
> (5) No protection from typos when reading env vars
> (6) Or when writing them
> (7) No fine-grained access control to env vars (occasionally useful if someone semi-technical wants direct control over something set in config but you don't want them touching the database credentials)
> (8) Depending on where you deploy to, your UI for reading and writing env vars may be better or worse
> (9) Depending on where you deploy to, your mechanism for cloning an environment's vars may be better or worse
> (10) No revision history ("Hey Patrick, can we go back to using that redirect we were using last week?")
> 
> For 1 - 5 I normally end up writing a wrapper around ENV with a little DSL to define the keys the app uses, so that I end up with file with a list of things like:
> 
> namespace :fulfilment_centre do
>   set :hours_of_operation, type: :integer, array: true, default: [9,17]
> end
> 
> And then I can call Config.fulfilment_centre.hours_of_operation with some confidence
> 
> But I don't really know how best to get round 6 - 10. I sort of feel like I want a Config-as-a-Service platform with web/RESTful interfaces for defining environments and their config, so that my apps can pull in config on deploy. Does something like that exist? Or (more likely, I would imagine) do I want the wrong thing? And if so why?
> 
> Any thoughts welcome!
> 
> Best,
> 
> Patrick
> _______________________________________________
> Chat mailing list
> Chat at lists.lrug.org
> Archives: http://lists.lrug.org/pipermail/chat-lrug.org
> Manage your subscription: http://lists.lrug.org/options.cgi/chat-lrug.org
> List info: http://lists.lrug.org/listinfo.cgi/chat-lrug.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lrug.org/pipermail/chat-lrug.org/attachments/20190227/e54b5b3f/attachment-0001.html>