You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Jackson 2.15 already has opt-in feature that improves writing of floating-point values (float, double).
Since Vector use case has to write large arrays of FP values this seems like a potential for improvements.
The only (?) question is whether writing of BigDecimal is improved as well; this is not guaranteed given different underlying number representation.
Turns out that writing (serialization) of BigDecimal only uses JDK and there is no optimized solution there (unlike with reads, see #653).
Will close this issue.
As per Blog post:
https://medium.com/@cowtowncoder/jackson-2-15-faster-floating-point-writes-too-19958e310185
Jackson 2.15 already has opt-in feature that improves writing of floating-point values (
float
,double
).Since Vector use case has to write large arrays of FP values this seems like a potential for improvements.
The only (?) question is whether writing of
BigDecimal
is improved as well; this is not guaranteed given different underlying number representation.Testing can be done by modifying one of benchmarks of https://github.com/FasterXML/jackson-benchmarks.
If this shows potential improvement, should also enable this setting for JSON API.
The text was updated successfully, but these errors were encountered: