Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This reverts commit f30dd0ff0cead9bb1f8f3b1b295392d75705486b.
|
|
|
|
|
|
|
|
|
|
Fix parsing sub-message field.
|
|
Added google/protobuf/descriptor.upb.* to core.
|
|
This is in preparation for making upb_def capable
of parsing binary descriptors directly.
We leave upb/descriptor/descriptor.upbdefs.* in
place for now, because upb/descriptor/reader.*
still depends on it.
Also removed a bit of cruft from the codegen.
|
|
Buffer start place needs to be adjusted before parsing sub-message.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
If the compiler elects to represent enum E as a char rather than an int
(per 6.7.2.2p4), the call to va_start() will have undefined behavior.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1. For decoding, an unknownfields will be lazily created on message,
which contains bytes of unknown fields.
2. For encoding, if the unknownfields is present on message, all bytes
contained in it will be serialized.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|