Ongoing developments

[pthom]

Hi @dcnieho :

I just opened this discussion so that we can exchange about ongoing development in a dedicated zone. I hope it will be OK for you.

I'll send the make ImVec2/4 iterable patch later tonight.

Superb!

[pthom]

I need to inform you about important and fruitful changes I just applied: see 1d98595

By changing the generator option options.fn_exclude_non_api = False (and also correcting the generator as a consequence 🫤), I added lots of new features, and they seem quite useful).
Look at the diff especially on bindings/imgui_bundle/implot.pyi and bindings/imgui_bundle/implot.pyi

You will see that we got ImVec2.__getitem__ for free, + ImLerp, ImClamp, Contains and other goodies.

This code now works:

    vec2 = ImVec2(1, 2)
    x = vec2[0]
    y = vec2[1]

ImVec2 is still not iterable, no __iter__ can be generated from C++.

[dcnieho]

I can do either, you may vote :)

[dcnieho]

(assuming both work)

[dcnieho]

I'll just do it on the C++ side, that works well

[pthom]

I can do either, you may vote :)

I vote for three lines of python if it works. It is much more maintainable.

[dcnieho]

Ah hmm, i just tunnel-focused on delivering what i set out to and went to bed, didn't see this until now. Hope the C++ PR is good for you too.

[dcnieho]

Cool! Will check the diff in detail, and see if i can make it iterable, may be just another missing function.

[pthom]

Also see 2a44816 which adds warnings for the final user.

[dcnieho]

very nice stuff!

[pthom]

You will be happy: 4f4d25e : ImGuiStyle.Colors is now bound as an python list/array

Beware, I forked imgui_toggle (and had to update it). You may need to git submodule sync

[dcnieho]

Super! will test right away

[dcnieho]

It doesn't work for me. If i set a color through the array, the value is not updated. The interface doesn't change color, and if i get the color again, it is not my new value.

But see my comment here, there is (perhaps) a more general way to fix this than a patch? #38 (reply in thread)

Not high priority.

It does make me wonder: does your generator output what things it doesn't wrap? Would be good to look through to see if some things should be wrapped, and if there are patterns/low hanging fruit where support for one case nets a lot (like the linked discussion)

[pthom]

It does make me wonder: does your generator output what things it doesn't wrap?

Yes it does

[pthom]

It doesn't work for me. If i set a color through the array, the value is not updated. The interface doesn't change color, and if i get the color again, it is not my new value.

This is strange. See http://www.giphy.com/gifs/IzoZ6z0xCGUgm16i8R

This is with the sandbox app: demos_python/sandbox_app.py

[pthom]

Oh, I get it this is when you change the whole ImVec4 at a time.

Well, this will be for another day. I'm gonna call it a day, and you should to :-) Thanks a lot.

[pthom]

Hi Dee,

I merged your to PRs this morning. Yesterday I took a day as I was very tired.

Also, see 567eb10: I removed the access to ImGuiStyle.colors as an array, as it was broken.
Temporarily, we haveImGuiStyle.color_(i) and ImGuiStyle.set_color_(i)`(note the trailing underscore, which leaves an open possibility for a port as an array of something close)

With your cast ImVec2 from list, imgui.color_edit_4 now works perfectly, see sandbox_app.py

[dcnieho]

Thanks for that, glad they were useful! Of course man, you have been making some late nights i saw. Do not feel pressure from me.

I saw that commit indeed, will give it a test soonish.

[pthom]

@dcnieho

Please see #52 where I opened a discussion concerning array-like access to C array members. There is an initial study to do about the feasibility if you feel like investigating :-)

Concerning the generator and your attempts to compile it, it was never compiled under windows. Before going further, I'd like to discuss this with you by telephone (and may be help a little by remote access). I could be available at 3PM or 5PM.

[dcnieho]

Hmm, nice start! Not quite there yet in that ideally the machinery would be invisible to the end user, who just see an array. I'd love your help getting the generator to work, but i can't today, will contact you over email.

[pthom]

@dcnieho : warning concerning the latest commits to the dev and main branch.

I reorganized all the external libraries folder structure: see 8673434

The external folder structure now looks like this:

• Each library original source is stored inside external/{lib_name}/{lib_name}
• The bindings scripts + cpp pydef are stored inside external/{lib_name}/bindings
• The generation scripts are stored inside external/bindings_generation

(The intent was to try to group the sources by library as much as possible).

Also see more comprehesive doc about bindings inside https://github.com/pthom/imgui_bundle/blob/main/external/_doc/Howto_Bindings.md

As a consequence, git will very probably choke when you run git update.

Your solutions are

• either to re-clone the repo from scratch
• or suppress external/ and .git/modules/external/ before updating the repo

(this will happen only once)

thanks!

[dcnieho]

Indeed, it choked. So i recloned, seemed easiest. Current dev HEAD built fine with pip install .

…

Message ID: <pthom/imgui_bundle/repo-discussions/49/comments/4576524@ github.com>