RFR: 8307132: Cleanup the code of sun.java2d.cmm.lcms package

[Phil Race]

On Thu, 4 May 2023 06:15:12 GMT, Sergey Bylokhov <serb at openjdk.org> wrote:

>> src/java.desktop/share/native/liblcms/LCMS.c line 51:
>> 
>>> 49: #define DT_BYTE     sun_java2d_cmm_lcms_LCMSImageLayout_DT_BYTE
>>> 50: #define DT_SHORT    sun_java2d_cmm_lcms_LCMSImageLayout_DT_SHORT
>>> 51: #define DT_INT      sun_java2d_cmm_lcms_LCMSImageLayout_DT_INT
>> 
>> I don't understand why this isn't a step backwards.
>> It is overloading the "size" to mean the "type"
>> How would you tell the difference between SHORT vs UNSIGNED SHORT - both 2 bytes,
>> or INT vs FLOAT - both 4 bytes ?
>> or LONG vs DOUBLE ..
>> If we ever need to, then this all needs to be reverted.
>> % cd java/lang 
>> %  grep BYTES *java 
>> Byte.java:    public static final int BYTES = SIZE / Byte.SIZE;
>> Character.java:    public static final int BYTES = SIZE / Byte.SIZE;
>> Double.java:    public static final int BYTES = SIZE / Byte.SIZE;
>> Float.java:    public static final int BYTES = SIZE / Byte.SIZE;
>> Integer.java:    public static final int BYTES = SIZE / Byte.SIZE;
>> Long.java:    public static final int BYTES = SIZE / Byte.SIZE;
>> Short.java:    public static final int BYTES = SIZE / Byte.SIZE;
>
> right, the idea was to use one variable to select the supported type using its size.
> Based on size we will switch what GetXXArrayElements to use. java right now supports these related functions:
> 
>     GetByteArrayElements,
>     GetShortArrayElements,
>     GetIntArrayElements,
>     GetLongArrayElements,
>     GetFloatArrayElements,
>     GetDoubleArrayElements,
> 
> The images based on long format currently are not supported by java2d and lcms. And the "float" formats are really slow in LCMS as mentioned in the description.
> 
>>How would you tell the difference between SHORT vs UNSIGNED SHORT - both 2 bytes,
> 
> Both will use the same code path.
> 
> But I can switch this to using two different variables for type and size.

I'd prefer that.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/13732#discussion_r1185263336