PDB Serialization (Pt. 2) #71
Conversation
src/pdb/mod.rs
Outdated
| test_roundtrip_with_args( | ||
| &[ | ||
| 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 47, 0, 0, 0, 32, 0, 0, 0, 0, 0, 0, 0, 7, 64, 0, |
There was a problem hiding this comment.
Where did you get this from? Just exporting from rekordbox?
There was a problem hiding this comment.
Yes, IIRC It's a page from the rekordbox export in this repo. I should've documented this in a comment.
There was a problem hiding this comment.
I'll add a comment when I open the PR with the solution.
Don't worrry and thanks for looking into this. |
|
I think I've gotten closer to the solution. One problem is that the export we generate has the rows in reverse. The other one is that rekordbox has additional padding between each row. I'll try to find a reason or a pattern to emulate. |
src/pdb/mod.rs
Outdated
| #[br(offset = page_heap_offset, args { count: num_rows.into(), inner: (page_type,) })] | ||
| rows: Vec<FilePtr16<Row>>, | ||
| #[br(offset = page_heap_offset, parse_with = FilePtr16::parse, args { count: num_rows.into(), inner: (page_type,) })] | ||
| rows: Vec<Row>, |
There was a problem hiding this comment.
This is not the same and thus this branch also broke plain reading unfortunately. The proper implementation is a vector of pointers while the new one is a pointer of a vector.
Also, since the manual write implementation RowGroup::write_options_and_get_row_offset is already quite complex I'm questioning it makes sense to continue down this path. In the end it comes down to requiring some sort of FilePtr serialization IMO. Also, some sort of abstract heap would be nice too (since Pages are also essentially just heap allocated too).
There was a problem hiding this comment.
Well, that would require computing the offsets beforehand (to instantiate FilePtr32), which would make it hard to create a PDB file from scratch. The user likely does not care about the offsets. This is why I chose this approach, but I'm open to ideas.
Holzhaus
force-pushed
the
pdb-serialization-pt2
branch
from
6a7e25d
to
0b11bcd
Compare
Holzhaus
force-pushed
the
pdb-serialization-pt2
branch
from
0b11bcd
to
3f90b88
Compare
|
Removed the broken code so that we can get the working changes in. Other stuff will go into a later PR. |
Currently not working correctly, as the unittest fails. I'm currently a bit busy and don't have too much time to debug it. If anyone else wants to start working on it or give feedback, I'd appreciate it.
Related issue: #68