[OpenJDK 2D-Dev] RFR: 8242004: TextLayout throws Exception with a non-invertible transform

[Philip Race]

On 4/9/20, 10:26 PM, Jayathirth D v wrote:
> Hi Phil,
>
> I went through all use cases captured in test case (TextLayout, drawXXXX).
>
> With updated change there is difference in behaviour between how we 
> interpret non-invertible transform between TextLayout.draw() and 
> drawXXXX() API’s.
> In case of TextLayout.draw() we are overriding non-invertible 
> transform and allowing text rendering to happen, but in case of 
> drawXXXX() we just return and doesn’t allow text rendering to 
> continue. Is it okay to have this difference in behaviour?

It becomes tricky.
Do you have a suggestion ?
Remember that the TextLayout is returned and does not have to be drawn, 
but could be
by both drawing it directly or asking for the outline shape and 
rendering that.
It can also be queried for the layout etc. There needs to be something 
returned that
does not cause other problems. And patently there can't be apps that 
would care because
today they can't get that far.
And there's no defined behaviour in this case.

So if you have specific code suggestions ..
>
> Also in test case its better if we continue to test all use cases and 
> then fail instead of failing at first instance and added test case 
> needs change in Copyright year from 2015 to 2020.

oops.

-phil.

>
> Thanks,
> Jay
>
>> On 10-Apr-2020, at 7:53 AM, Philip Race <philip.race at oracle.com 
>> <mailto:philip.race at oracle.com>> wrote:
>>
>> D**n copy/paste, yes you correctly inferred the webrev is at
>> <cr-url>/<my openjdk id>/<bugid> ie : 
>> http://cr.openjdk.java.net/~prr/8242004/ 
>> <http://cr.openjdk.java.net/%7Eprr/8242004/>
>>
>> -phil.
>>
>>
>> On 4/9/20, 7:00 PM, Jayathirth D v wrote:
>>> Hi Phil,
>>>
>>> Please share webrev link, you have added JBS link for webrev.
>>> I went to path where you usually share webrev's and found 
>>> http://cr.openjdk.java.net/~prr/8242004/ 
>>> <http://cr.openjdk.java.net/%7Eprr/8242004/>
>>>
>>> Thanks,
>>> Jay
>>>
>>>
>>>> On 10-Apr-2020, at 12:49 AM, Philip Race <philip.race at oracle.com 
>>>> <mailto:philip.race at oracle.com>> wrote:
>>>>
>>>> Any takers ?
>>>>
>>>> -phil
>>>>
>>>> On 4/3/20, 1:29 PM, Philip Race wrote:
>>>>> bug: https://bugs.openjdk.java.net/browse/JDK-8242004
>>>>> webrev: https://bugs.openjdk.java.net/browse/JDK-8242004
>>>>>
>>>>> Several code paths can end up in the method shown in the bug report
>>>>> with a non-invertible transform.
>>>>>
>>>>> As much as possible, we can prevent them reaching here by checking 
>>>>> in the rendering code.
>>>>>
>>>>> If we do get here, which should now be possible only when directly 
>>>>> creating
>>>>> a TextLayout, we can use a default TX. Subsequent rendering of the 
>>>>> TextLayout
>>>>> will be handled by the other checks being added.
>>>>>
>>>>> A regression test is provided which checks various APIs to make 
>>>>> sure no
>>>>> exception is thrown.
>>>>>
>>>>> -phil.
>>>>>
>>>>>
>>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/2d-dev/attachments/20200410/2219ae59/attachment-0001.htm>