[MDEV-23762] Strange type-inference/rounding problem since >10.3.23 - Jira

XML

Word

Printable

Details

Type: Bug
Status: Closed (View Workflow)
Priority: Critical
Resolution: Duplicate
Affects Version/s: 10.1.46, 10.3.24, 10.4.14, 10.5.5
Fix Version/s: N/A
Component/s: Optimizer
Labels:
- regression
Environment:
CentOS7

Description

Hi,
we have some problems with "POW()" or the inference of the term: (1/(1+(0.0275/12)).
Since upgrading to any version above 10.3.23, the statement

SELECT POW((1/(1+(0.0275/12))), 59)

yield complete interesting (and wrong) results.

SELECT VERSION(), POW((1/(1+(0.0275/12))), 59), POW((1/(1+(CAST(0.0275/12 as double)))), 59), POW((1.0/(1.0+(0.0275/12.0))), 59)

produces:

# VERSION(), POW((1/(1+(0.0275/12))), 59), POW((1/(1+(CAST(0.0275/12 as double)))), 59), POW((1.0/(1.0+(0.0275/12.0))), 59)

'10.3.23-MariaDB', '0.8736690781345217', '0.8736690781345217', '0.8736690438487815'

'10.5.5-MariaDB', '0.8729680736160053', '0.8736688724200776', '0.8734844622108874'

Can you explain/confirm/fix this issue? The first Column was our usecase and differs a lot between these versions!

Greetings,
Robin

Attachments

Issue Links

is caused by

MDEV-19232 Floating point precision / value comparison problem

Closed

is duplicated by

MDEV-23702 calculating(auto rounding) issue

Closed

links to

regression commit 62d73df6b270cc94ba577e96d3bf325170f306fe

Activity

People

Assignee:: Varun Gupta (Inactive)

Reporter:: Robin Parker

Votes:: 0 Vote for this issue

Watchers:: 3 Start watching this issue

Dates

Created:: 2020-09-18 22:24

Updated:: 2020-09-19 22:34

Resolved:: 2020-09-19 08:22

Git Integration

Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.