> I assume the test fails without the source change.
Yes, it fails without the fix.
Also I should mention I ran all our Image I/O regression tests and they
all pass.
Of course only some fraction of those test the GIF decoding path.
-phil.
On 2/12/2020 11:55 AM, Brian Burkhalter wrote:
On Feb 12, 2020, at 11:25 AM, Phil Race <philip.r...@oracle.com
<mailto:philip.r...@oracle.com>> wrote:
Bug:https://bugs.openjdk.java.net/browse/JDK-8238842
Webrev:http://cr.openjdk.java.net/~prr/8238842/
The string table has a maximum size of 4096 - by GIF spec as well as
in the implementation
The value initCodeSize (as a power of 2) to initialise the string
table, and since GIF
only allows 8bpp, this ought to max out at 256.
But there's no check so we can get an ArrayIndexOutOfBoundsException
This is now checked for.
Looks good.
Also I removed a printStackTrace which briefly confused me during
debugging
as I had no idea where the message was coming from.
I saw.
See the bug for more details.
I assume the test fails without the source change.
Brian
