Maciej Pilichowski wrote:

Hi,

But if there is no reference (no previously selected region for example) it would be useful to paste the block initially at mouse cursor position (counting left, upper corner of the rectangle which boundaries of the block make, as the base point).

I don't think it is a good idea to use the cursor as the insertion point since this is both very uncommon and not very practical, the mouse cursor is too volatile for this. More reasonable is to use the selection as the insertion point, and this is currently how it works. Using this as a base for an improved workflow seems like a better idea to me.

/ Martin

On Thu, 2009-05-14 at 18:43 +0200, Martin Nordholts wrote: [...]

I don't think it is a good idea to use the cursor as the insertion point since this is both very uncommon and not very practical, the mouse cursor is too volatile for this. More reasonable is to use the selection as the insertion point, and this is currently how it works. Using this as a base for an improved workflow seems like a better idea to me.

When there is no selection, and you paste, the paste typically ends up 3,926,201 screens above where you are working (for me at least). So the workflow is
1) paste
2) zoom out until yu find the selection 3) when you can't see it, turn on the marching ants 4) use the move tool and shift-arrow to drag the selection (or drag with the mouse if it's solid so you can actually grab it when it's only a few mm long on the screen) 5) zoom in on the place where you want to work, a step at a time, gradually moving the floating selection 6) when you get to 50% or 100% so you can work, try to remember why you wanted whatever you pasted.

Why you think that's a smoother workflow than 1) paste so floating selection appears on the screen 2) continue working

is beyond me.

When there *is* a selection, normal procedure is 1) undo the paste
2) select none
3) paste and follow the 6-step path to confusion given above :-)

For me at least.

Liam

Maciej Pilichowski wrote:

So, I opt for some kind of predictability vs. randomness. And in such case, mouse cursor position is a good reference.

I don't think the cursor is ever a good insertion point. It is better than a random insertion point, yes, but so what? It's still not a good insertion point IMO.

/ Martin

Liam R E Quin writes:

5) zoom in on the place where you want to work, a step at a time, gradually moving the floating selection 6) when you get to 50% or 100% so you can work, try to remember why you wanted whatever you pasted.

Why you think that's a smoother workflow than 1) paste so floating selection appears on the screen 2) continue working

is beyond me.

I've found the "paste centers" behavior quite useful, and have recommended it to lots of other people as a quick way to center a layer (which used to be a FAQ, though less so now that the align tool exists).

That said, it's hard to see an argument for anyone wanting to paste to a part of the image that isn't visible. Surely most people paste because they want to do something with the pasted layer, which requires it being visible on screen.

Perhaps paste should center within the visible viewport?

...Akkana

Hi,

On Thu, 2009-05-14 at 11:17 -0700, Akkana Peck wrote:

I've found the "paste centers" behavior quite useful, and have recommended it to lots of other people as a quick way to center a layer (which used to be a FAQ, though less so now that the align tool exists).

We could add "Center Layer" to the menus.

Sven

Hi,

On Thu, 2009-05-14 at 20:48 +0200, Maciej Pilichowski wrote:

On Thursday 14 May 2009 20:17:43 Akkana Peck wrote:

I've found the "paste centers" behavior quite useful,

It is predictable and more useful than random placement for sure. But with hires monitor I would still like some kind of hint from the mouse. Maybe LMB click and then paste would do it?

So I am for dropping random placement, and using centered placement.

GIMP doesn't place the pasted content randomly. What makes you think so?

Sven

Hi,

On Thu, 2009-05-14 at 13:21 -0400, Liam R E Quin wrote:

When there is no selection, and you paste, the paste typically ends up 3,926,201 screens above where you are working (for me at least).

Not sure what version of GIMP you are using. But the current code has the following logic:

If there is a selection, paste to the center of the selection boundary.

If there is no selection, paste to the center of the viewport unless that would place the selection completely outside the drawable we we are pasting to.

If we still didn't find a suitable position, fall back to the center of the image.

I didn't check all corner cases, but the code looks OK and the little tests I did seem to indicate that the behavior is as I described.

Sven

When you paste a second time, the first paste should still be visible and selected(?) and the floating selection is the current drawable, and thus the second paste end up on top of it (allowing for difference in paste sizes), right? Can you elaborate on the precise order of operations that results in the second paste ending up somewhere "random"?

On Fri, May 15, 2009 at 3:16 PM, Maciej Pilichowski wrote:

On Friday 15 May 2009 20:40:21 Sven Neumann wrote:

GIMP doesn't place the pasted content randomly. What makes you think so?

Because I don't see any relevance in second paste to what I do (and where I do) and I see no relevance between first paste and the second one. And it should be.

Hi,

On Fri, 2009-05-15 at 21:16 +0200, Maciej Pilichowski wrote:

GIMP doesn't place the pasted content randomly. What makes you think so?

Because I don't see any relevance in second paste to what I do (and where I do) and I see no relevance between first paste and the second one. And it should be.

I explained the currently implemented logic in another mail. It is by far not random. Did you even try a recent development snapshot before you posted your wishes here?

Sven

Hi,

On Fri, 2009-05-15 at 21:42 +0200, Maciej Pilichowski wrote:

On Friday 15 May 2009 21:23:57 Sparr wrote:

When you paste a second time, the first paste should still be visible and selected(?) and the floating selection is the current drawable, and thus the second paste end up on top of it (allowing for difference in paste sizes), right? Can you elaborate on the precise order of operations that results in the second paste ending up somewhere "random"?

As Sven explained it is centered indeed, not truly random. But to realize the need of sane placement of the second paste, take an image, zoom it in, copy a rectangle from left, upper corner, paste it, confirm, paste it again, you have to go for the block (to d&d) right, down. Ok drag it and drop it in the initial corner. Now, copy something from bottom, right corner, paste it, confirm, paste it again. Now you have to go up, left.

Simple enough to press Shift-Ctrl-A if you don't need your selection any longer. Than what you paste won't end up being placed there.

Sure, perhaps this can still be improved, but it needs careful thinking and a proper analysis of the current behavior and possible work-flows. A lot of thought and effort has gone into the current behavior. And I refuse to discuss such changes with someone who completely disrespects this effort and calls the placement random.

Sven

Logic may be even changed, but before change something is better understand how it works

Paste is centered at the center of the selection (if any)

Is not simple and intuitive create, or move, the selection where you want your object be paste ?

How i would able to guess a relation with the position of my mouse cursor and a object to be pasted ? i can not see any logical connection

But i may well guess a relation with the selection

Only other meaningful relation i am able to imagine is center at the intersect point of 2 guides IF snap to guide is enabled (and no more then 2 guides are used)

If not will be required for GIMP not only develop a advanced AI but even telepathic skills , because if for you may be relevant the position of the mouse, for me may be much more relevant the center of the portion of image i zoomed in...and who know what may seems more relevant for somebody else

Maciej Pilichowski wrote:

So my workflow would be like this (option b): * copy
* paste
* confirm
* paste
* (this will move the block from its initial position to the position of the mouse)

ctrl+shift+h is of course just an example. Sticky dragging is similar, it "glues" the block to the mouse, so you don't need to hold LMB while dragging (relief for users with even RSI).

User in such case could choose whatever fit the best current situation.

What do you think?

I think the concept your propose is interesting, but I also feel it is highly optimized for your own workflow, and doing interaction design based on several peoples personal desires is never a good idea. I'd like to see your proposal presented in terms of the bigger GIMP UI picture. This site will be a starting point for doing such a thing: http://gui.gimp.org/index.php/GIMP_UI_Redesign

/ Martin