Gavin King ( https://hibernate.atlassian.net/secure/ViewProfile.jspa?accountId=557058%... ) *commented* on HHH-16650 ( https://hibernate.atlassian.net/browse/HHH-16650?atlOrigin=eyJpIjoiMDU3MT... ) Re: Loss in precision due usage of Float data type instead of BigDecimal in case Oracle JDBC driver returning -127 for scale and 0 for precision if precision/scale is unknown. Might happen unexpectedly and requires very close inspection of used SQL ( https://hibernate.atlassian.net/browse/HHH-16650?atlOrigin=eyJpIjoiMDU3MT... ) OK, thanks, so, conclusion: * This only happens in quite specific circumstances (selecting numeric literals, it seems) where for some reason Oracle is unable to assign a proper type to the literal, and it reports the scale as -127. It does not appear to occur when just selecting columns. * This tripped up our workaround for a different bug where Oracle never correctly reports the type of FLOAT / DOUBLE PRECISION columns. We were using a scale of -127 to detect the FLOAT type, since Oracle always incorrectly reports it as NUMBER. But this turns out to be fragile (no surprise). * And, so, in cases of types with more than 15 digits of precision, this could lead to truncation of the least-significant digits. Now, it’s worth keeping in mind that more than 15 decimal digits of precision is incredibly rare in real world data. My proposed fix is here: https://github.com/hibernate/hibernate-orm/pull/6613/commits/55bc38dede84... ( https://hibernate.atlassian.net/browse/HHH-16650#add-comment?atlOrigin=ey... ) Add Comment ( https://hibernate.atlassian.net/browse/HHH-16650#add-comment?atlOrigin=ey... ) Get Jira notifications on your phone! Download the Jira Cloud app for Android ( https://play.google.com/store/apps/details?id=com.atlassian.android.jira.... ) or iOS ( https://itunes.apple.com/app/apple-store/id1006972087?pt=696495&ct=Em... ) This message was sent by Atlassian Jira (v1001.0.0-SNAPSHOT#100225- sha1:9dd72a2 )