Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
User-facing change (minor)
The outputs archives for transforms and synthetics (which are singleton objects we treat sort of like "buckets" holding all the outputs from multiple
run_transforms
/generate
calls) are now "archives of archives" instead of archives of files organized in subdirectories:(where "t1" and "t2" are run identifiers—can be user supplied but default to current timestamp)
Performance details
The previous approach to creating regular archive files (everything used a single
add_to_tar
function) was woefully inefficient. In the case of archives that are only made once with no nesting (e.g. source data, synthetics training datasets) we now havearchive_items
. Using the same set of non-trivial-sized tables mentioned in #117 :In the case of the "bucket archives" with results from multiple runs described above (run transforms multiple times, generate multiple times), the new
archive_nested_dir
is significantly better as more and more runs are added:Individual times for each operation: