Yes, good point - I'll make the member final. And also re-apply my patch
combined w/ yours.

Thanks
Shai

On Tue, Nov 30, 2010 at 9:31 AM, Uwe Schindler <[email protected]> wrote:

> --- From: Shai Erera [mailto:[email protected]] ---
> Uwe, I'm sorry people rushed in to believe I broke backwards ... because I
> don't think I did. My only mistake was that I didn't run test-backwards,
> 'cause I didn't really think anything can break by using CHM and not HM.
>
> fileMap was package-private until two days ago (LUCENE-2778) when I made it
> protected - therefore it wasn't a public API, right? If it wasn't public,
> we're allowed to change it right? The break was only in MockRAMDir, and
> even
> that is because I changed fileMap type from HashMap to Map, which IMO
> should
> have been defined like that from the beginning.
> ---
>
> Hi Shai,
>
> Sorry I did not know that this field was package private before, I was only
> looking at the test. I am then fine with the patch. I have seen the other
> mail about backwards breaks and backwards tests - here the fix for this
> stupid HashMap/Map issue is easy, simply apply the BW patch attached to the
> issue.
>
> One thing: As you made the map protected in the previous issue, and it is
> never changed, for API safety it should be final in all cases! Can you
> change that, too. The reference to the Map should never be changed.
>
> See also my comments to the other mail!
>
> Uwe
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>

Reply via email to