Why is there a "dynamic inline table" when you normally want a static type to do dumping?

[goodboy]

After much pain and anguish i'd sure like to know why this code exists.

It entirely breaks your encoding lookup table (if you try to use it) and just generally makes all of your .dump_sections() code more or less an adhoc mess for anyone that wants to write "inline tables" (aka in #384, #396).

I can't seem to find much argument in any of the encoder code for why there is this weird "special encoders" stuff either, when most of the overloads are just for stuff that could be covered by the TomlEncoder.dump_funcs system anyway.. or at the least pick whether you want flags or subclassing..

Seems like there's a whole lot of design bypassing and just general strange choices for the lack-of-symmetry between the encoding and decoding system.

[goodboy]

lol, i mean this docstring isn't even anywhere near right. what's the reason it isn't already inheriting from dict..i'd sure love to know 😂