Fork me on GitHub
#malli
<
2019-12-09
>
rschmukler17:12:23

@ikitommi just confirming that with the changes, a custom :string/decode on a schema entry will still fully overwrite the default mt/string-transformer for the type, yes? I think that escape hatch is quite useful.

ikitommi17:12:59

@rschmukler yes, will override, no changes there

rschmukler17:12:21

Wonderful! Reading through the code but it looks great so far

ikitommi17:12:47

just chains all the different names.

rschmukler17:12:02

Hmmmm - do we even need transformer-chain protocol method?

rschmukler17:12:05

Is it just for introspection?

rschmukler17:12:42

I guess we're using it to assemble the interceptor chain

ikitommi17:12:43

oh, good point.

rschmukler17:12:10

But I think we could move all of that into the transformer function and then simplify Transformer to a single value-transformer method, which could be nice

rschmukler17:12:30

It ended up being critical for my application (the ability to encode the renaming of keys for future transformers) but I'm open to alternative approaches if you've got other ideas you want me to explore

rschmukler17:12:51

Part of me thinks using metadata is odd - part of me thinks that this is the perfect use case for it

rschmukler17:12:09

ie. it is metadata about how the new value was created... so maybe it makes sense?

rschmukler17:12:54

One thing we may have to do though, if we end up sticking with that API, is merge the metadata about transformers as we compose them

rschmukler17:12:22

Also, random aside, but malli is Finish, correct?