wilson chen

@vapier , hi, i added `--enable-cxx` option in my last commit. is it what we want? please have a review. thx

> > integrating libgd with c++ without any option sounds resonable considering the relation between c and c++. > > sorry but I don't know what you mean by relation....

i think it is reasonable. what about adding it to GD 2.3 or 2.4? @vapier

similar to #344

> An update: Google has just announced another feature in Google Images that will (usually) be driven by IPTC XMP metadata. ("Licensable", they're calling it.) They already support three copyright-related...

Hi @carlseibert3, Sorry for my late reply. The support for metadata is great feature. I am a new maintainer to this community, but I promise I will try my best...

it is easy to implement 5x5 matric convolution. We could add an api like `gdImageConvolution5x5()`. But do should we construct different 5x5 filter to implement `gdImageEdgeDetectQuick`, `gdImageGauseeianBlur`, `gdImageEmboss`, `gdImageMeanRemoval`, `gdImageSmooth`?

Hi @fila43, thanks for your feedback. Here is your patch: ``` Fix problems with testing against installed GD library --- libgd-gd-2.3.0/CMakeLists.txt 2020-03-22 14:40:02.000000000 +0100 +++ ../libgd-gd-2.3.0/CMakeLists.txt 2020-08-10 11:04:59.670292207 +0200 @@...

not only `gdImagePng`, but the following, `gdImageGif`, `gdImageWBMP`, `gdImageJpeg`, `gdImageWebp`, `gdImageTiff`, are unable to detect failure. changing their return value type to `int` is fine, but that will course api...

I think the key problem is whether libgd should deal with non utf-8 character. If yes, we should improve `gdImageCreateFromFile(const char *filename)`, or we need a new API with input...