[
https://issues.apache.org/jira/browse/LUCENE-3281?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13060786#comment-13060786
]
Robert Ragno commented on LUCENE-3281:
--------------------------------------
Ah, good point. It seems cleaner to maintain it (which is straightforward). The
other sensible alternative would be to make the asserts all refer to the
up-rounded capacity. However, it seems reasonable and consistent to have an OBS
present the capacity it was constructed with.
I suppose there is room to split capacity() and size(), but that might confuse
existing uses.
Incidentally, if it were open to behavioral changes... I would find it more
convenient if the asserts were replaced with assuming that the vector was
infinite, filled with zeros. This seems more consistent with the set
operations, anyway. And the union operation, and so on. (And it is not as if
anyone can properly be relying on the current asserts to control flow.)
> OpenBitSet should report the configured capacity/size
> -----------------------------------------------------
>
> Key: LUCENE-3281
> URL: https://issues.apache.org/jira/browse/LUCENE-3281
> Project: Lucene - Java
> Issue Type: Bug
> Components: core/other
> Affects Versions: 3.0, 3.0.1, 3.0.2, 3.0.3, 3.1, 3.2
> Reporter: Robert Ragno
> Priority: Minor
> Original Estimate: 2m
> Remaining Estimate: 2m
>
> OpenBitSet rounds up the capacity() to the next multiple of 64 from what was
> specified. However, this is particularly damaging with the new asserts, which
> trigger when anything above the specified capacity is used as an index. The
> simple fix is to return numBits for capacity().
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
