updated code to purescript 0.12 and Generic.Rep

[CarstenKoenig]

ok this includes a lot of (sadly breaking) changes:

• updated packages to the newest ones for PureScript 0.12
• uses purescript-argonaut-generic (right now my own fork - I hope they include this or something similar soon)
• uses Aff
• I think it's better to have a version of affjax (called it ajax here) that is close to your original version but there is one change: it will use argonaut-generic instances to decode the JSON right there - so we don't need getResult any more - I think it's the obvious use case here
• it also drops most of the copied code and uses affjax directly - only FFI left is encodeUriComponent and unsafeToString - mostly because I could not find the first any more and I'm not sure what the second is exactly used for at the moment

sadly there are no tests and I'm not 100% certain that this code is working

so this is more or less WIP but I need this to continue on servant-purescript and I wanted to have your input/feedback on that

best regards,
Carsten

[eskimor]

why is this special case no longer needed?

[CarstenKoenig]

if I did not mess up this should be taken care of by this instance here:

https://github.com/eskimor/purescript-servant-support/pull/15/files#diff-7b8d38eeda45bc94e96d07a922b4631bR37

as far as I understood your code this one was used pass on strings as they are

[eskimor]

haha - yeah I completely missed that one. Looks good then. 'else instance' - that's a nice construct :-P

[eskimor]

I believe the second is used for error reporting, if I remember correctly.

If we get the JSON directly, how do we report errors if something goes wrong?

[CarstenKoenig]

I should probably write some test cases but I hoped to use this kind of error

https://github.com/eskimor/purescript-servant-support/pull/15/files#diff-4dd6048806498efe169a5f3a62c9672bR59

for those kinds of errors

(argonaut-generic: https://github.com/CarstenKoenig/purescript-argonaut-generic/blob/RowToList/src/Data/Argonaut/Decode/Generic/Rep.purs should produce hopefully readable errors for why some JSON could not be decoded)

[eskimor]

Sounds reasonable.

[shmish111]

ok, this one is needed, what is the situation with this, I will test with the servant-purescript PR now

[vlatkoB]

Any updates on the update? :-)

[legrostdg]

any news?