Rename RemoveIgnored -> RemoveMembers #249
Conversation
ajor
force-pushed
the
rename-removed-ignored
branch
4 times, most recently
from
deaddf0
to
49cc599
Compare
Also reshuffle CodeGen's passes to fix an alignment bug with removed members. Change RemoveMembers to actually remove members instead of replacing them with padding. AddPadding must be run afterwards to fill in the gaps.
ajor
force-pushed
the
rename-removed-ignored
branch
from
49cc599
to
6aad43a
Compare
| Member{paddingArray, AddPadding::MemberPrefix, c.members[i].bitOffset}; | ||
| } | ||
| std::erase_if(c.members, [this, &c](Member member) { | ||
| return ignoreMember(c.name(), member.name); |
There was a problem hiding this comment.
Given c.name() returns a copy, can you gather it before and capture it by reference?
| } | ||
|
|
||
| bool RemoveIgnored::ignoreMember(const std::string& typeName, | ||
| bool RemoveMembers::ignoreMember(const std::string& typeName, |
There was a problem hiding this comment.
Unrelated but nearby, why is this not checking a set? It seems better to implement a hash function for std::pair<std::string, std::string> than loop over a vector of pairs. Even a map of ContainerName -> Set would be a lot better.
There was a problem hiding this comment.
The real reason: It's sharing the data structure with CodeGen v1 and I didn't want to change too much.
But in terms of efficiency, I think the winner of vectors vs maps/sets would come down to how many elements are typically being stored. For a small number of elements a vector would likely win due to the lack of hashing and being more cache friendly.
Also reshuffle CodeGen's passes to fix an alignment bug with removed members.
Change RemoveMembers to actually remove members instead of replacing them with padding. AddPadding must be run afterwards to fill in the gaps.