-
Notifications
You must be signed in to change notification settings - Fork 65
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix chunking issues in NDFileHDF5 #404
Conversation
…reen for additional chunking dimensions
…reen for additional chunking dimensions
…umExtraDims=0, which is incorrect
Someone who uses extraDimensions should check that it is still working correctly after the chunking changes. The unit tests in pluginTests do some extra dimension tests, and they are passing OK, but someone should do a more real-world test. |
@ajgdls will take a look at this one |
I've run a suite of system tests that check the multi-dimensional writes, positional placement of arrays within extra dimensions and chunking. They all pass except for two tests, which involve different chunking values when using extra dimensions. I believe these two do not fail as a result of your changes here @MarkRivers but instead fail due to an attempt to incorrectly write them with direct chunk write. However I'll push the fix to this branch if you don't have any objections as it is related. |
That sounds good to me. |
…at the TIFF standard allows. Allow use of TIFFTAG_LAST_ATTRIBUTE tag. Previous it only allowed up to TIFFTAG_LAST_ATTRIBUTE-1. Fixed error in definition of NUM_CUSTOM_TIFF_TAGS which could lead to access violation.
Hi @MarkRivers @GDYendell I have pushed a commit with the extra check to ensure direct chunk write isn't used if extra dimension chunk sizes are not 1. All of our system tests and the unit tests now pass. I would appreciate if you could take a look at this commit as the change was not quite as simple as I had anticipated. |
It looks OK to me. Am I right in thinking this isn't supposed to pass the extra chunk dims through, it just checks they are 1 here and then blindly inserts 1s later when using direct chunk write? |
I didn't alter any of the code that actually performs the writing (as that was already working as expected), so I guess the answer to your question is yes this is purely checking the values here to decide if we can direct chunk write or not. |
Fix chunking issues in NDFileHDF5
This PR fixes the issues identified in #401: