[
https://issues.apache.org/jira/browse/MAHOUT-337?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12845519#action_12845519
]
Jake Mannix commented on MAHOUT-337:
------------------------------------
Depending on how Pallavi was seeing this come up (doing surely not
String.equalsIgnoreCase() on the json strings? Ordering is not guaranteed on
the key-value pairs of the sparse outputs, for one thing), there are a couple
of solutions. The "right" way to tell if two vectors are the same is to do
AbstractVector.equivalent(Vector). We should make that more clear in the
documentation, certainly.
Actually, now that I read the implementation of equals() in the vector classes,
they're *horribly* inefficient (don't take into account sparsity at all), and I
really hope it's not being used anywhere that performance matters (but for the
record: equals and equivalent correctly ignore the cached value when comparing
vectors)
Should the string form of the vector have the cached value? It's actually
useful knowledge to debug with, as I think I mentioned in the list thread - if
you expect that this value should not be changing, or if you need a quick
"mental checksum" that this is the vector you're expecting to see, keeping
track of that one double value is a good signature of what the vector is.
I guess I'm in favor of "not-a-bug" and just documenting it. But I'm open to
further thinking on how to make it easy to keep track of this kind of thing.
> Don't serialize cached length squared in JSON vector representation
> -------------------------------------------------------------------
>
> Key: MAHOUT-337
> URL: https://issues.apache.org/jira/browse/MAHOUT-337
> Project: Mahout
> Issue Type: Bug
> Components: Math
> Affects Versions: 0.3
> Reporter: Sean Owen
> Assignee: Sean Owen
> Priority: Minor
> Fix For: 0.4
>
>
> The cached length-squared field in vectors should be marked transient so that
> it is not part of the JSON serialized state.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
