[
https://issues.apache.org/jira/browse/LUCENE-5759?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Adrien Grand updated LUCENE-5759:
---------------------------------
Attachment: LUCENE-5759.patch
New iteration. I tried to decouple the rounding logic from the computation of
the number of bits required in DirectWriter. Consumers first need to call
bitsRequired or unsignedBitsRequuired depending on their needs, and then to
pass the result to DirectWriter.roundBits to get an amount of bits per value
that is supported by the writer.
> Add PackedInts.unsignedBitsRequired
> -----------------------------------
>
> Key: LUCENE-5759
> URL: https://issues.apache.org/jira/browse/LUCENE-5759
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Assignee: Adrien Grand
> Priority: Minor
> Fix For: 4.9
>
> Attachments: LUCENE-5759.patch, LUCENE-5759.patch
>
>
> Across the code base, we have lots of:
> {code}
> long minValue, maxValue;
> final long delta = maxValue - minValue;
> final int bitsRequired = delta < 0 64 : Packedints.bitsRequired(delta);
> {code}
> {{Packedints.bitsRequired(delta)}} doesn't work directly in that case since
> it expects a positive value. And that is important that it does so in order
> to get an error instead of silently being super wasteful if a negative value
> is provided.
> Yet in some cases such as the one depicted above, the value should be
> interpreted as an unsigned long. So I propose to add another {{bitsRequired}}
> method that would interpret the value as unsigned.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
