Add anonymous datasets #917
Conversation
src/HDF5.jl
Outdated
| dapl::DatasetAccessProperties = DatasetAccessProperties(), | ||
| pv... | ||
| ) | ||
| path !== nothing && haskey(parent, path) && error("cannot create dataset: object \"", path, "\" already exists at ", name(parent)) |
There was a problem hiding this comment.
Could you clarify what you would like to test? I already test passing nothing to the function.
The code for checking for an existing dataset already existed before this PR. Is there currently not a test for this?
There was a problem hiding this comment.
I think we already test the error for an existing dataset:
Lines 1362 to 1367 in 704d9e8
There was a problem hiding this comment.
Nevermind. When I read this, I thought you added this whole line, but that's not right heh
src/HDF5.jl
Outdated
| pv = setproperties!(dcpl,dxpl,dapl; pv...) | ||
| isempty(pv) || error("invalid keyword options") | ||
| Dataset(API.h5d_create(parent, path, dtype, dspace, _link_properties(path), dcpl, dapl), file(parent), dxpl) | ||
| if path isa Nothing |
There was a problem hiding this comment.
| if path isa Nothing | |
| if isnothing(path) |
would be more canonical
There was a problem hiding this comment.
https://docs.julialang.org/en/v1/base/base/#Base.isnothing is only available with Julia 1.1. Have we broken compat with Julia 1.0?
There was a problem hiding this comment.
|
Interesting feature, didn't know this was possible. Apparently the python docs suggest this can be helpful for linking later. I still kind of wonder what kind of use case that would assist. |
Removed space within Union. Used `=== nothing` for consistency and Julia 1.0 compat. Co-authored-by: Mustafa M <mus-m@outlook.com>
src/HDF5.jl
Outdated
| dtype::Datatype, | ||
| dspace::Dataspace; | ||
| dcpl::DatasetCreateProperties = DatasetCreateProperties(), | ||
| dxpl::DatasetTransferProperties = DatasetTransferProperties(), | ||
| dapl::DatasetAccessProperties = DatasetAccessProperties(), | ||
| pv... | ||
| ) | ||
| haskey(parent, path) && error("cannot create dataset: object \"", path, "\" already exists at ", name(parent)) | ||
| path !== nothing && haskey(parent, path) && error("cannot create dataset: object \"", path, "\" already exists at ", name(parent)) |
There was a problem hiding this comment.
| path !== nothing && haskey(parent, path) && error("cannot create dataset: object \"", path, "\" already exists at ", name(parent)) | |
| isnothing(path) || haskey(parent, path) && error("cannot create dataset: object \"", path, "\" already exists at ", name(parent)) |
src/HDF5.jl
Outdated
| pv = setproperties!(dcpl,dxpl,dapl; pv...) | ||
| isempty(pv) || error("invalid keyword options") | ||
| Dataset(API.h5d_create(parent, path, dtype, dspace, _link_properties(path), dcpl, dapl), file(parent), dxpl) | ||
| if path === nothing |
There was a problem hiding this comment.
| if path === nothing | |
| if isnothing(path) |
Co-authored-by: Mustafa M <mus-m@outlook.com>
|
In my application I'm using this to create a binary template for LabView to fill in with data. The anonymous datasets allows me to space out the real datasets at regular intervals so it is easy to fill in. This way LabView just knows that needs to know to write dataset at regular offsets and it can "create" a HDF5 file without using the HDF5 library. |
This adds
HDF5.API.h5d_create_anonto allow for the creation of anonymous datasets.If
nothingis passed as the path or name tocreate_datasetor related functions, then an anonymous dataset is created. This mirrors what happens ifNoneis passed to h5py: https://docs.h5py.org/en/stable/high/group.html#h5py.Group.create_dataset