Open theotherphil opened 5 years ago
I'd make this slightly dependent on the image-core
rework of @fintelia. In its essence, imageproc
could be based on image-core
such that it is largely independent of image
. For the current moment I'd leave everything as is and at some point begin the process of migrating functions over. What's your impression of the stability and maturity of imageproc
in comparison to image
?
Sorry, missed your comment at the time. Depending on a new image-core sounds good to me.
imageproc
is definitely less stable and less mature than image
as a whole, but as mature as the imageops
module in image
.
If I may I'd like to weigh in on this as having as little image processing stuff as possible part of the image
crate. I'd prefer to have it concerned with reading and writing image formats only. Image processing generally doesn't depend on the on-disk image format at all, so having it be separate makes sense.
Both the
image
andimageproc
crates contain functions for performing (some) affine transformations, colour and contrast adjustments, and image filtering.Does anyone have a strong opinion on which functions should live where, or on an appropriate crate structure for the union of the functions provided by this crate and
imageproc
?I have no preference for the final crate structure, but I would like to get rid of the current duplication of functionality, and would ideally like the resulting split to be easily discoverable.