Codecov Report

Merging #8 into master will increase coverage by 20.81%.
The diff coverage is 90.90%.

@@              Coverage Diff              @@
##             master       #8       +/-   ##
=============================================
+ Coverage      0.00%   20.81%   +20.81%     
- Complexity       78       82        +4     
=============================================
  Files             1        1               
  Lines           188      197        +9     
=============================================
+ Hits              0       41       +41     
+ Misses          188      156       -32     

The 07d1608 commit fixes a stupid mistake, and should be squashed into 19cd533.

Thanks again Guy!

Two thoughts..

1. I've never seen any algorithm use a default weight of 50.. It's usually 0 or 100..
2. Would it be an idea to add support for a sorting callback-method instead, so people can have their own sorting algorithm?

Regardless of the above, this needs a unit test..

1. I've never seen any algorithm use a default weight of 50.. It's usually 0 or 100..

I chose 50 to try and match what users of SSP might expect. The authproc filters set an implicit scale of 0-100, with 50 being used as the "default" in the documentation. Nevertheless, am not wedded to it, and am happy to change it to 100.

2. Would it be an idea to add support for a sorting callback-method instead, so people can have their own sorting algorithm?

Hmm. A hook. I'll have a look at this.

Even so, I think that weights are still useful as a simple mechanism. If I have four social providers I want to order (as I do), I don't want to have to write a hook to fix the order; weights in the metadata config are much easier.

Regardless of the above, this needs a unit test..

I knew you'd say that, and I'd have done so if there were already unit tests for PowerIdPDisco.php. I think this is a bigger problem than just this change. IMHO some documentation is just as important, perhaps more so.

Yeah, unit testing is a bigger problem.. But you gotta start somewhere, so that's why I asked :)
We're putting a lot of work in it lately and managed to get the saml2-lib to ~90% 🎉

Woah, that's great work Guy! 🚀

After giving @tvdijen's callback idea some thought, it seems the easiest and most flexible way to do this is simply to call a hook if it exists. If we use the existing hook functionality, that requires passing the entire metadata array to the hook. However, that in turn gives the user/admin complete control of the sorting including using functions other than uasort.

I think I'm all done here now :)

Thanks Guy!!