[
https://issues.apache.org/jira/browse/THRIFT-964?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12924361#action_12924361
]
Arvind Jayaprakash commented on THRIFT-964:
-------------------------------------------
Mathias, it should be possible to take the underlying 4 bytes that represent a
single precision number and shove it into some other data type of a similar
size. Two problems exist with this approach
(1) The need for a custom serialization/deserialization will end up making it
an "out of trunk" feature against which the library users will have to code
their application.
(2) Space+time inefficiency.
Outside of C (& ASM), I do not know of any language where a contiguous region
of memory can be reinterpreted as an array of another type. As I mentioned,
size a reason for asking this feature in the first place. If I deserialize the
4 bytes as anything but a float and then assign it to a float, it will be a
value copy in most languages. If I had to go through such memory overheads, I
might as well stick to double.
> Support for single precision floating point numbers
> ---------------------------------------------------
>
> Key: THRIFT-964
> URL: https://issues.apache.org/jira/browse/THRIFT-964
> Project: Thrift
> Issue Type: New Feature
> Components: Compiler (General)
> Reporter: Arvind Jayaprakash
>
> We have an application that uses "float", the IEEE single precision numbers.
> The reasons are the two obvious ones: the lack of need for precision as high
> as "double" and data size given that we have millions of such numbers.
> I'd like to see support for this data format.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
