Mark1626 opened a new pull request, #18999:
URL: https://github.com/apache/datafusion/pull/18999

   ## Which issue does this PR close?
   
   - Closes #17555 .
   
   ## Rationale for this change
   
   ### Analysis 
   
   Other engines:
   1. Clickhouse seems to only consider `"(U)Int*", "Float*", "Decimal*"`  as 
arguments for log 
https://github.com/ClickHouse/ClickHouse/blob/master/src/Functions/log.cpp#L47-L63
   
   Libraries
   
   1. There a C++ library libdecimal which internally uses [Intel Decimal 
Floating Point 
Library](https://www.intel.com/content/www/us/en/developer/articles/tool/intel-decimal-floating-point-math-library.html)
 for it's 
[decimal32](https://github.com/GaryHughes/stddecimal/blob/main/libdecimal/decimal_cmath.cpp#L150-L159)
 operations. Intel's library itself converts the decimal32 to double and calls 
`log`. 
https://github.com/karlorz/IntelRDFPMathLib20U2/blob/main/LIBRARY/src/bid32_log.c
   2. There was another C++ library based on IBM's decimal decNumber library 
https://github.com/semihc/CppDecimal . This one's implementation of 
[`log`](https://github.com/semihc/CppDecimal/blob/main/src/decNumber.c#L1384-L1518)
 is fully using decimal, but I don't think this would be very performant way to 
do this
   
   I'm going to go with an approach similar to the one inside Intel's decimal 
library. To begin with the `decimal32 -> double` is done by a simple scaling
   
   ## What changes are included in this PR?
   
   1. Support Decimal32 for log
   
   ## Are these changes tested?
   
   Yes, unit tests have been added, and I've tested this from the datafusion 
cli for Decimal32
   
   ```
   > select log(2.0, arrow_cast(12345.67, 'Decimal32(9, 2)'));
   +-----------------------------------------------------------------------+
   | log(Float64(2),arrow_cast(Float64(12345.67),Utf8("Decimal32(9, 2)"))) |
   +-----------------------------------------------------------------------+
   | 13.591717513271785                                                    |
   +-----------------------------------------------------------------------+
   1 row(s) fetched. 
   Elapsed 0.021 seconds.
   ```
   
   ## Are there any user-facing changes?
   
   1. The precision of the result for Decimal32 will change, the precision loss 
in #18524 does not occur in this PR
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to