And again I fall into trap of checking performance of functions with
empty data set. I used notes that I generated during org-roam v1 and
they had invalid format of tags, meaning that there were 0 notes in
the resulting data set, meaning that I didn't really test for notes
retrieval.
Now that is fixed by the following steps.
Test set was generated again with org-roam v2 and it contains 9554
notes with links, properties and proper tags.
To make sure that the data set is not malformed, there is a special
check that result of benchmarked function is not empty.
Last, but the thing that actually fix performance - bulk retrieve
all matched notes instead of using many calls to
vulpea-db-get-by-id. This means that we have one more HUGE and
FRIGHTENING SQL query, but seems like performance is much better.
See updated README file.
And again I fall into trap of checking performance of functions with empty data set. I used notes that I generated during org-roam v1 and they had invalid format of tags, meaning that there were 0 notes in the resulting data set, meaning that I didn't really test for notes retrieval.
Now that is fixed by the following steps.
Test set was generated again with org-roam v2 and it contains 9554 notes with links, properties and proper tags.
To make sure that the data set is not malformed, there is a special check that result of benchmarked function is not empty.
Last, but the thing that actually fix performance - bulk retrieve all matched notes instead of using many calls to
vulpea-db-get-by-id
. This means that we have one more HUGE and FRIGHTENING SQL query, but seems like performance is much better. See updated README file.