[OpenJDK 2D-Dev] RFR: 8166038 BufferedImage methods getTileGridXOffset() and getTileGridYOffset() return a non 0 value for sub images

[Martin Desruisseaux]

Hello Jay

Le 17/04/2020 à 12:58, Jayathirth D v a écrit :

> I have a question regarding test case. Why test has to check whether 
> we are getting negative value or not and then calculate whether the 
> tileGridOffset is 0. From different function's specification in 
> BufferedImage I see that there are no multiple tiles support. I went 
> through mail thread related to how we are comparing it to JAI’s 
> implementation. But I think we can still keep the test case to be 
> minimal and just check for 0 grid offset values.
>
It is true that just testing that grid offset values are 0 would be 
sufficient. But doing so does not explain why it needs to be zero 
(except because the specification said so). The purpose of the first 
part of the test, which is computing tileX and tileY, is to illustrate 
why grid offsets need to be zero. The idea is to said "Suppose that we 
don't know that this image has only one tile. Let's calculate tile 
indices the way we would do for an arbitrary RenderedImage. Do we get 
consistent results?".

In other words the two tests are redundant. But the first test requiring 
(tileX == 0 && tileY == 0) takes the perspective of checking if the 
values are mathematically consistent, while the other tests requiring 
(tileGridXOffset == 0 && tileGridXOffset == 0) are basically the same 
tests but from the perspective of checking compliance with 
specification. The first test about mathematical consistency is a kind 
of documentation, but addresses the problem which was the reason for 
this patch request.

The comment about negative value is because the formulas used in the 
test — tileX = (x - tileGridXOffset) / image.getTileWidth() — is not 
quite exact. The division should be replaced by a call to 
Math.floorDiv(int, int). I though that it was not necessary for this 
test since the two formulas will differ only if (x - tileGridXOffset) is 
negative, which is not the case in this test, and that a comment would 
be simpler. But maybe it would be less confusing to remove the comment 
and use Math.floorDiv(int, int) instead.

     Regards,

         Martin


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/2d-dev/attachments/20200417/0c0cc42c/attachment.htm>