Comment 1:

Owner changed to @nigeltao.

Comment 2:

That play.golang.org program gives the raw bytes; can you describe how that test image
was generated?

Comment 3 by jsummers3456:

I don't know, I think I modified one of my programs that uses libpng, then used
<https://github.com/jteeuwen/go-bindata>;. The goal was to make a minimal testcase
for this bug.
Any non-paletted image with a tRNS chunk will do. There are examples in PngSuite
<http://www.schaik.com/pngsuite/PngSuite-2011apr25.tgz>;, such as tbrn2c08.png.

Comment 4 by Vova616:

Looks like it makes problems after trying to read "gAMA" chunk which isn't supported yet.

Comment 5:

Labels changed: added size-m.

Comment 6:

Labels changed: added suggested.

Comment 7:

Labels changed: added priority-later, removed priority-triage.

Status changed to Accepted.

Comment 8:

The provided sample image is not a valid PNG image.
The mentioned tbrn2c08.png of
http://www.schaik.com/pngsuite/PngSuite-2011apr25.tgz
is also invalid.
The spec is pretty clear about tRNS has to come after PLTE and
before IDAT.  See:
http://www.libpng.org/pub/png/spec/iso/index-object.html#5ChunkOrdering
It seems as if this was not check in libpng prior to 1.6.0
but is enforced now.
See
http://sourceforge.net/p/libpng/code/ci/0da9cf38cd55178e6dabacf720fa21741bc50cd5/tree/pngrutil.c#l1001
So: Not a bug.

Comment 9:

Labels changed: removed go1.1, size-m, suggested.

Status changed to WorkingAsIntended.