bungeman

Taking a quick look at [producer_qtext.cpp](https://github.com/mltframework/mlt/blob/8f2108ee6d0d5849484f2f54de3d7e7adfc4aba7/src/modules/qt/producer_qtext.cpp), it looks like the strategy is to retrieve the path of the text and then draw the path. However, Noto Color Emoji is bitmap...

> In the TTX format there is now an "axis name" which maps to an "axis ID", which I find a little confusing. Yeah, that seems strange. The issue is...

> really the axes don't have ids, the ids are owned by the fvar table and the order in which the axes are laid out there. Just to add to...

Since I just noticed it, in the `` table XML serialization the `` has its own `` with an `index` attribute. So it would at least be consistent to instead...

There is no reason for '#' to imply a hidden axis. One would hope that, as much as possible, the xml format should try to be a higher level view...

Calling `hb_face_count` on the blob first is needed if the user wants to know if `hb_face_create` will recognize the data or not. If the data is not recognized `hb_face_create` will...

When this is implemented, note that [icc_4.ksy](https://github.com/kaitai-io/kaitai_struct_formats/blob/09f41727eacbe1ab453dc6d2fdb9a66199c6d2c6/image/icc_4.ksy#L1211) needs an (integer) exponent operator to calculate the size of a multi-dimensional array. Currently this is incorrectly using the `^` bitwise xor operator...

Hmmm... it looks like `generateDocumentOutline` ended up in devtools-protocol [r1226504](https://github.com/ChromeDevTools/devtools-protocol/commit/9a978928587a2ba68d479fbfc0be735802a5450d) but it looks like devtools-protocol is still pinned to `r1213968`. So this will need to wait for a devtools-protocol roll....

Did you happen to see what happened in`SkCTFontDescriptorGetSkFontStyle`? There are a few things that could happen here, some of which are fixable on the Skia side and some which aren't....

That is unfortunate. I had been hoping that the data was present but for some reason in an unexpected format, but if we're not getting the data at all then...