RFR: 8377568: DataBuffer constructors and methods do not specify required exceptions [v4]

[Phil Race]

On Wed, 18 Feb 2026 18:47:03 GMT, Phil Race <prr at openjdk.org> wrote:

>>> You get a valid DataBuffer but there isn't a normal use for it.,
>> 
>> Yes, calling any methods on it will cause an exception, right? In that case, we should probably reject it? Or are there cases where an empty data buffer could actually be useful?
>
> I can't think of any use for it once it is created.
> 
> Because everything that uses a DataBuffer (in the imaging APIs) like Raster and BufferedImage require w/h of at least 1. So I'm not sure how a zero-length DataBuffer could ever be used once created.
> And getElem() / setElem() can never be used either. 
> The only use  would be in tests that see if you can create it  .. not a particularly practical real world use.
> 
> But because the spec. didn't disallow it, I didn't want to do anything about it. I tried to keep this change as much as possible to enforcing what the existing spec says. Already a tiny bit worried about breaking some weird code somewhere that ignores the spec.

I slept on this and I can't think of any legitimate use that needs a DataBuffer of size 0. Hope I'm right on this, since I've updated the code and the spec to disallow zero. Pushing that to the PR in just a few moments.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/29766#discussion_r2830977478