Support for directly reading HDF5 and NetCDF MINC files. #296
Conversation
…ix V1 headers with V2 headers.
…sumes nonexistant padding at EOF.
hdf5 improvment
| dv_offset = new_off; | ||
| } | ||
| switch (typ) { | ||
| case type_enum.INT8: |
There was a problem hiding this comment.
Nitpick. Can you properly indent the case statement ? Or you indent like that due to jshint ?
There was a problem hiding this comment.
I generally follow the common convention of keeping the "case" clauses at the same indentation as the "switch" statement. The brainbrowser codebase is inconsistent here - some of the existing code indents cases, some does not.
|
@rdvincent I read all your code. Make some little comments. Nice job, it will be really useful.
|
|
@natacha-beck I will try to formalize some of the comments to conform to docular. I don't really know the docular standard, but I'll try to follow it where appropriate. Up until now I was mostly just trying to leave comments for myself. |
2 participants
Implement #272. This is still something of a work in progress, but it has reached the point where I believe it will read most MINC 1.0 and MINC 2.0 files correctly. Ideally there should be more testing, enhancement, and documentation before any actual release incorporating this change. But I would like to get comments and review started.
The HDF5 code is large and hairy, which reflects the size and complexity of the specification. HDF5 files incorporate two different classes of objects, and the different classes can coexist in one file. The format is full of interdependencies between different structures, making abstraction a challenge.
In comparison, the NetCDF "classic" format is simple, so the code is much easier to write and read.