With 7.4.2, you can set COVERAGE_CORE=sysmon globally on your CI, and it'll only use it where available (Python 3.12 and 3.13 alpha), and use the default for 3.11 and older.
The recent news that MoneyDashboard is suddenly shutting down has exposed a gap in the way OpenBanking works. It is simply impossible for a user to get read-only access to their own data without using an aggregator. And there are very few aggregators around.
Why is it impossible for me to get programmatic access to my own data?
There are two interlinked reasons which I'd like to discuss.
OpenBanking is a brilliant idea encoded in an excellent standard wrapped in some very complex processes and with some rather unfair limitations.
OpenBanking presents a standardised API to allow read and write access to a financial account. So I could give a smartphone app read-only access to my credit card and let it automatically tell me when I've spent more than £50 on sausage rolls this week. Or I could add all my bank accounts to one service which would let me see my net worth. Or any of a hundred ideas.
I could also connect my accounts in such a way that when Bank Account A drop below £100, an OpenBanking API request is sent to Bank Account B to transfer some money to A.
But here's the first problem. The only way you can get access to a bank's API is if you have a licence. And you only get a licence if you're a financial institution who can prove that they have robust security controls. Which means that individuals have to go through an aggregator. Or, in OpenBanking terms, an "Account Information Service Provider".
Some OpenBanking providers will let individuals play in a "sandbox" to test out the API. There are no real accounts and no real money, it's just a way to test how the API works.
I can see why that makes sense for write access. You don't want a user's unpatched Raspberry Pi suddenly sending all their money to Russia.
And I can see why that makes sense for organisations which deal with data from multiple people. One leak and everyone is exposed.
But I'm not convinced that it makes sense to deny an individual read-only API access to their own account. Sure, I might accidentally leak my own data - but the same risk exists if I download a PDF statement from my bank.
The second problem is that not every OpenBanking consumer will talk to every OpenBanking provider.
For example, I have an account with Coventry Building society. They have an OpenBanking API which no one uses! They're not the largest financial institution in the UK, but have a fair few customers. And yet all the OpenBanking apps refuse to work with it.
So even if I did find an aggregator with an API, it may not work with all my financial institutions.
"Every #newsroom in every community needs to think about #ClimateChange not as a beat but as a through line involving everything we do. No corner of the newsroom is exempt—not business or culture, not sports or city hall.
On the national level, journalism has to figure out how to make climate change central to our #politics#coverage."
As a pretty new #django learner(1+ yr), I've set myself up the task of re-writing the main apps as function-based views. Most tutorials start with CBVs, so I wanted to 'unravel' them. Additionally, I'm re-writing the tests for #pytest, getting rid of SimpleTestCase & TestCase.
Just finished re-writing the 1st of my projects' app & got 100% ✅ test #coverage 🥳 It's really been helping me learn a great deal. It's a good struggle.
How do you organise your method handlers? Do you have separate functions for GET, POST, etc. or do you check request.method a bunch of times in a single function?
I like how CBVs allow you to group a view’s method handlers.
I find myself using CBGVs less and less as I can’t remember how things are called, what I need to overwrite, manually instantiate, etc.
Should we use #coverage reports only for #unittest or also for integration and functional testing?
Should the test directory split into unit, integration and functional tests or should all tests in one root tests folder? (Both following the namespace rules)
as for the directories: yes, I would split up the different types of tests into separate directories, and also into different test suites, so you can run them independently.
coverage is a useful tool for all tests, but has less value for functional or integration tests imho. it is mostly a great way to figure out which parts of your code are not tested yet. As such, you can use coverage on all three types, but seperately, not combined.
Eine Einführung in die Messung von Antennenanlagen und in die Bewertung, Analyse und Vermeidung passiver Intermodulationen (PIM) mit spezieller Betrachtung verteilter Antennensysteme (DAS) und den Einflüssen von PIM auf verschiedenen Mobilfunkbändern