[
https://issues.apache.org/jira/browse/LUCENE-7361?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15354281#comment-15354281
]
David Smiley commented on LUCENE-7361:
--------------------------------------
MemoryIndex is doing this itself. And if not this issue then in a separate
issue I think we should improve it (using the WIP code attached here as a
straw-man) so that it simply uses the Lucene public
Terms/TermsEnum/PostingsEnum APIs rather than needlessly using its internal
implementation data which led to a bug. Terms is Terms... why should
MemoryIndex be different? The only thing different I see is that MemoryIndex
is going to be not as huge as a main index (in general) so, subjectively, it
can sorta get away with overriding toString vs some other method. I'd actually
rather it didn't -- leaving term & position details be on another method to
avoid toString() getting humungous. Where I'm coming at this is that it'd be a
shame if this debugging utility only existed on MemoryIndex since the code
doesn't really care about MemoryIndex specifics; MI has no specifics -- it's an
index, albeit a small one.
At times I've wished to view an index I'm debugging that I'm writing tests for,
which has a small amount of data as it's the unit/test data. SimpleTextCodec
is one option but it's very inconvenient to switch to and switch to a non-RAM
directory vs. a hypothetical diagnostic method on Fields / Terms when I'm
already in the debugger poking around. Doesn't that seem useful to you too?
> Terms.toStringDebug
> -------------------
>
> Key: LUCENE-7361
> URL: https://issues.apache.org/jira/browse/LUCENE-7361
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: David Smiley
> Attachments: MemoryIndexToString.java
>
>
> While fixing LUCENE-7340, MemoryIndex.toString(), I thought MemoryIndex
> shouldn't need it's own debug toString() impl for its Terms when there could
> be a generic one. So here I propose that we create a
> Terms.toStringDebug(Appendable result, int charLimit, String indent) or
> some-such but probably not override toString() for obvious reasons. Maybe
> also have this on Fields() that simply loops and calls out to the one on
> Terms.
> The format is debatable.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
