tip with extra checks is good, deprecate even better, I will update patch....
----- Original Message ----
From: Michael McCandless <[EMAIL PROTECTED]>
To: java-dev@lucene.apache.org
Sent: Tuesday, 11 March, 2008 2:45:56 PM
Subject: Re: Ideas to refactor Filed
Hello!
Responses below:
eks dev wrote:
> Moin Moin Michael,
>
> for the first issue I have crated LUCENE-1217, and for the second
> one I have some questions.
>
> if we maintain length and offset internally in Field than we have
> one, imo, theoretical "legacy performance problem" as we need to
> create new byte[length] and copy in order to preserve compatibility
> (users expect this method to return compact array with 0 offset)
> I am talking about.
> public byte[] binaryValue();
Actually, if offset==0 and dataLength==array.length, can't we return
the array itself? This way legacy apps, which will pass both these
checks, would see tiny (because of these added checks) performance
loss? Also, in a search setting, where doc was created from stored
fields, I think both those checks would be true as well (unless
FieldsReader is changed to share byte[] arrays between fields).
I think we should then deprecate binaryValue() in favor of
getBinaryValue()?
> would that be acceptable, it is very small penalty and there will
> be a way to avoid it? Anyhow, if one is using
> public void setValue(byte[] value), it is to be expected that this
> user allready has a reference to value. This makes this
> question rather theoretical, no?
>
> we could than create new methods, getOffset() getLength()
> getBinaryValue() that enable full spectrum and replace all uses
> that expect 0-offset array.
Mike
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
__________________________________________________________
Sent from Yahoo! Mail.
The World's Favourite Email http://uk.docs.yahoo.com/nowyoucan.html
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
