On Mon, 27 Jun 2022 20:27:01 GMT, Sergey Bylokhov <[email protected]> wrote:

>> I did comment on this in the bug and the PR description.
>> Other exception types are possible or very difficult to prove they won't be 
>> thrown anprop if it ever did happen
>> we'd be back here. So this seems like the best to thing to do since we do 
>> not want to propagate them.
>
>> Other exception types are possible or very difficult to prove they won't be 
>> thrown anprop if it ever did happen
> we'd be back here
> 
> But then it will mean that the "read" thrown wrong exceptions, Isn't this 
> just hiding the actual bug?

In the end, the ImageIO method we change is just a wrapper for getting and 
using the reader, and that unexpected exception could be thrown if the readers 
are used directly via a similar public API.

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

PR: https://git.openjdk.org/jdk/pull/9113

Reply via email to