Support (de)serialize_with in tuples #335
Conversation
| @@ -177,31 +177,29 @@ pub struct FieldAttrs { | |||
| name: Name, | |||
| skip_serializing_field: bool, | |||
| skip_deserializing_field: bool, | |||
| skip_serializing_field_if: Option<P<ast::Expr>>, | |||
| skip_serializing_if: Option<ast::Path>, | |||
There was a problem hiding this comment.
iirc the idea was to allow arbitrary expressions in the future
There was a problem hiding this comment.
I see. Given that it's not on our roadmap (no issue for it) and I am not convinced that it would be a valuable thing to support, I do no think we should keep around a lot of convoluted dead code in the hope that some day it will be useful.
|
I rebased to fix the merge conflict with #338. We both added something to the bottom of serde_tests/tests/test_gen.rs. |
|
The travis failure is clippy doesn't build. |
|
I think I grokked most of the code now.
agreed on both points. We should probably add cfail tests some day to test for behavior we don't want. @homu r+ |
Support (de)serialize_with in tuples Fixes #330 and a number of related bugs around (de)serialize_with in newtype structs, tuple structs, newtype variants, and tuple variants. The most interesting thing in this PR is a significantly condensed version of `wrap_serialize_with`. It now works a lot more like `wrap_deserialize_with`, with just a thin wrapper that provides a `serialize` method. Please confirm that the contortions in the original implementation were not necessary. I had to make this change to support serialize_with in newtype structs and newtype variants.
|
hmmm right... well... it looks like clippy is going to work again soon, I'll postpone instead of merging manually |
I filed #341 to add cfail tests. The clippy PR we need to watch is https://github.com/Manishearth/rust-clippy/pull/944, going to be published as 0.0.69. |
|
@dtolnay done |
|
Merging because I retriggered the Travis build and it passed. |
Fixes #330 and a number of related bugs around (de)serialize_with in newtype structs, tuple structs, newtype variants, and tuple variants.
The most interesting thing in this PR is a significantly condensed version of
wrap_serialize_with. It now works a lot more likewrap_deserialize_with, with just a thin wrapper that provides aserializemethod. Please confirm that the contortions in the original implementation were not necessary. I had to make this change to support serialize_with in newtype structs and newtype variants.