Refl #10
Refl #10
Conversation
dimitrivlachos
force-pushed
the
refl
branch
2 times, most recently
from
0d161f8 to
5e62e92
Compare
| // Get the size of the datatype | ||
| size_t size = H5Tget_size(type_id); | ||
|
|
||
| // Match to supported types |
There was a problem hiding this comment.
(This is looking really nice btw!) I was giving this branch a test on a dials strong.refl (h5 format), and we also need to be able to handle arrays which are 64bit unsigned int (e.g. flags, panel). I am currently reading these like
std::vector<std::size_t> flags = read_array_from_h5_file<std::size_t>(.....).
There was a problem hiding this comment.
Thank you! I am quite happy with how much neater this implementation is 😁
Yes, I am aware that not all data types are being handled yet, I will sort that out.
| // Mark the group as visited | ||
| traverse_data->visited_groups->insert(full_path); | ||
|
|
||
| // Open the group to recurse into it |
There was a problem hiding this comment.
Following on from my previous comment, you may run into an issue when trying to open data from the shoebox group within the main data group, as there is a name clash (two 'panel' arrays). I think for now we should skip reading any data from any subgroups within the main data group as I don't expect we will be writing any for our new programs, or creating any shoebox-like objects from this data (feel free to leave the relevant code in but commented out).
There was a problem hiding this comment.
Ah, okay, I see. I should really index by the internal h5 path then. I will have a think about this.
There was a problem hiding this comment.
I don't think there is any need for indexing by the path, I think we just want to read the arrays in one given group only, not including subgroups, therefore there shouldn't be any name clashes?
There was a problem hiding this comment.
That would make sense 🤔
No description provided.