keyboard shortcuts for new points in cp editor panel

[wirz]

Dear list, especially Thomas,

Since some time in the last few weeks something is broken in the hugin
control points editor panel:

Assuming I have auto-add and auto-finetune turned off, if I click in one
image and the other image, I get a pair of 'new' points. It used to be
possible to press 'a' to add this pair. Now this doesn't do anything,
but I can press the 'add' button in the lower right corner. Now it used
to be possible to finetune the point positions of this new pair by
pressing 'f', however, this doesn't do anything any more. I can,
however, press the finetune button, or alternatively select a different
pair, and reselect the newly added pair: then pressing 'f' works.
Additionally, while a point pair is in a 'new' state, or has been added
but not yet de+reselected, zooming via '0', '1', '2' doesn't work.

I assume this isn't intentional? Is it obvious where this behaviour
comes from or shall I have a go at bisecting it (it must be very recent)?

cheers, lukas wirz

[wirz]

I quickly had a look what is going on: Commit 8697 introduces the
described behaviour, and more specifically, SetFocus() needs to be
called on mouseEnter. I do, however, not know how to fix this
appropriately. (There is also one superfluous semicolon.)

cheers, lukas wirz

[T. Modes]

Hi Lukas

lukas wirz schrieb am Sonntag, 16. Juni 2024 um 23:58:08 UTC+2:

I quickly had a look what is going on: Commit 8697 introduces the
described behaviour, and more specifically, SetFocus() needs to be
called on mouseEnter.

We can't call SetFocus unconditionally. This results in focus stealing issues.

I fixed this in an other way.

Thomas

PS: Had you time to look at the enblend bug again? The last report at https://groups.google.com/g/hugin-ptx/c/pRti1q4kiFs indicates that your fix is not enough.

[wirz]

Hi Thomas,

> We can't call SetFocus unconditionally. This results in focus stealing
> issues.
> I fixed this in an other way.

Awesome, thank you!

> PS: Had you time to look at the enblend bug again? The last report at
> https://groups.google.com/g/hugin-ptx/c/pRti1q4kiFs indicates that your fix
> is not enough.

I had seen the blended image in question, but didn't investigate
anything. There are several artefacts, a large area that is too bright,
a grey-ish edge, and something that might be a white/very bright
triangle, but nothing like the black corners we've seen before. I'd
have to spend a rainy Sunday on that some time.

cheers, lukas wirz