The problem I have with GIMP is that, aside from the insanely awful mass of windows spewed across the screen
Strange. I only get three windows on startup. One with the tool box and the tool settings, one with the layers, channels, paths and undo history, and one with the image (or, if no image is opened, a small empty area replacing that image). Actually, in the case where no image is open I'd like to get rid of the latter (just as it was with earlier versions). But since they also seem to have removed the menu from the tools window, I guess that's not possible (but maybe I've just not found the right option).
But at least we get "Merge the cage transform tool from GSoC" in this release. So now an arbitrary vector shape can be deformed and its containing pixels will deform naturally with it.
But we'll have to wait for version 3.8 before we can do rotation, translation and scaling in the same operation.
Priorities - The GIMP team has them.
Yeah, because it makes so much sense not to use some code some friendly person has written for you, and which does something which was impossible up to now, just because some other, unrelated feature which would give more convenience for doing things already possible now isn't finished yet.
Replying to fix an accidental "redundant" moderation (and killing four other moderations in the same step... oh for the good old days when moderating on Slashdot was a two-step process where you could easily correct a mis-click like this...)
At the moment, I duplicate (retype) any source material and provide a link to it.
And that is how it should be. If the cited text changes, you do not want your text, which refers to the old version, to suddenly apparently refer to the new version.
For example, say that Random Person writes on his web page "XY is a great man." Then you write on your web page "Random Person says: "XY is a great man." I'm inclined to agree." Now Random Person decides to change its mind and replaces that text with "XY is a complete asshole." Now you probably didn't change your mind, but in Ted Nelson's scheme your web page would now suddenly read "Random Person says: "XY is a complete asshole." I'm inclined to agree."
Not really. As the rest of the world moves to IPv6, the US will eventually have to follow even while having IPv4 addresses available, or it will lose accessibility to a growing part of the net. And the rest of the world will have a head start on IPv6, due to being forced to.
In my view, the ideal interface would use three mouse buttons:
Left mouse button: Select. This includes selecting menu items etc. Middle mouse button: Modify (move, rotate,...) Right mouse button: Context menu.
Of course there's the problem that with scroll wheel mouses, the middle button is combined with the scroll wheel, which makes that interface harder to use than with an old non-scroll three-button mouse. And of course, laptops almost always lack a middle button.
Do they? I would say, the learning curve describes the learning effort which has to be invested over the mastery to be achieved. A steep learning curve means a lot of learning for a little achievement. A barely noticeable incline would mean you are able to achieve great things with only a little learning.
Learning over time doesn't tell you anything about how difficult something is; it only tells you how hard-working or lazy the learner is.
Until Wings3D can actually be controlled with the keyboard instead of context menus alone, it won't be fast to use.
Of course, adding keyboard shortcuts to an UI is trivial without otherwise changing the UI. Therefore simply add "with keyboard shortcuts added" to the OPs statement to deal with that.
Until it can have more then one level of subsurf preview it won't be effective to.
I admit I have no idea what a subsurf preview is. However, is the limitation you mention really a limitation of the UI, instead of a limitation of the functionality? That is, would it be possible to add that functionality without changing the UI (apart from possibly adding another option somewhere)?
I have no idea what the Wings3D UI looks like, therefore I cannot say if it is really that good, but it seems to me that you didn't mention anything which would not be easily added when copying the UI.
I've never failed to find an unused spot (and some programs also hide the mouse pointer when working with the keyboard). Most programs have white space in the menu bar, many programs have scroll bars where you can put them, etc.
Moreover, I'm using "lazy focus-follows-mouse" where a window doesn't lose focus just because the mouse leaves it, but only when it enters another window. When I have non-maximized windows, I almost always have some unused desktop space around it where I can move the mouse cursor.
But then, most of the time I don't even bother to move the mouse cursor away completely. For example, as I write this, it's still in the input field. Only if it happens to disturb me at the place where it currently is, I move it away.
The main problem I have with non-mouse based focus is that the focused window always has to be on the top. I frequently put helper programs on top of the main program, at places where they don't disturb, and I hate it if they go away (hidden by the main program) every time I work with the main program. Now logically it would be possible to decouple keyboard focus and raised window even for click to focus, but nobody does (probably for the simple reason that those who dislike raise on click are those who prefer focus-follows-mouse anyway, so there would be no users using that feature).
Then you're emulating MDI through the window manager
No, MDI is (poorly) emulating a window manager. Unlike with MDI, on a WM workspace I can group windows from several different programs I use together. With MDI, this is not possible.
I don't have a problem with a program offering MDI for those poor people who cannot use a real window manager. However I have an issue with programs forcing an MDI on me, when my window manager is orders of magnitude better than the emulation the MDI program forces on me (partially by the mere fact that the MDI by design is limited to one application, and partially by the fact that I put considerable effort in the window manager to behave the way I like it, and of course the MDI application will ignore my window manager's settings, and that's of course ultimately also inherent, because no one can expect an application to know how emulate arbitrary window managers).
From other comments, it seems that GIMP did it the right way (let's just hope that they will never remove the "antiquated" independent-window interface for the sake of simplifying their code).
It's ideal if you never type errors, but I suspect with equality checking vs assignment, there's always going to be some mistakes since "A equals B" is how we talk about both equality and assignment.
You say "a equals b" when you talk about assignment!? I definitely don't.
But a solution where imperfect humans are less likely to make errors to begin with is better than one where they are likely to make errors. Of course, the compiler should catch the error in both cases.
How do mod points work now? I haven't had any mod points (that I know of) since the redesign.
That's funny. I've gotten lots of mod points lately, definitely more than in earlier times. Maybe it's related to me using all of them. Because otherwise, the only thing I do is post comments (I didn't metamoderate a single post for quite some time, and after a few failed attempts I've given up on trying to post articles).
Not really. Object variables are references, and those references are passed by value. It may not be the way you normally think about it, but it is technically correct. You never pass objects in Java, neither per value or per reference. Of course, passing a reference by value is equivalent to passing the referenced object by reference, and generally you'll think about it in exactly that way, but that doesn't make his statement wrong.
Kill off interfaces (even James Gosling admits that they were probably a mistake), add multiple inheritance.
Up to hear I'm with you (well, not completely; instead of killing interfaces, they could be turned into something useful).
A c/c++ style macro compiler and #include system
A definite NO! If there's anything bad about C/C++, it's the preprocessor. Nothing against macros, but please not C/C++ style. And C/C++-style #include isn't a good idea at all; the only reason there's not yet a sane alternative in the C++ standard is that it wasn't considered ready.
Slashdot Mirror
Strange. I only get three windows on startup. One with the tool box and the tool settings, one with the layers, channels, paths and undo history, and one with the image (or, if no image is opened, a small empty area replacing that image). Actually, in the case where no image is open I'd like to get rid of the latter (just as it was with earlier versions). But since they also seem to have removed the menu from the tools window, I guess that's not possible (but maybe I've just not found the right option).
Nope - a unified transform tool is slated for The GIMP 3.8.
http://wiki.gimp.org/index.php/GIMP_Roadmap
But at least we get "Merge the cage transform tool from GSoC" in this release.
So now an arbitrary vector shape can be deformed and its containing pixels will deform naturally with it.
But we'll have to wait for version 3.8 before we can do rotation, translation and scaling in the same operation.
Priorities - The GIMP team has them.
Yeah, because it makes so much sense not to use some code some friendly person has written for you, and which does something which was impossible up to now, just because some other, unrelated feature which would give more convenience for doing things already possible now isn't finished yet.
I guess the original meaning is this one.
The number of atoms in the universe would conveniently fit into your garage as well. It only has about 80 digits.
Replying to fix an accidental "redundant" moderation (and killing four other moderations in the same step ... oh for the good old days when moderating on Slashdot was a two-step process where you could easily correct a mis-click like this ...)
Simply add a run time limit for the link testing. If the link target cannot be checked after a short time, then the link is suspicious anyway.
And that is how it should be. If the cited text changes, you do not want your text, which refers to the old version, to suddenly apparently refer to the new version.
For example, say that Random Person writes on his web page "XY is a great man." Then you write on your web page "Random Person says: "XY is a great man." I'm inclined to agree." Now Random Person decides to change its mind and replaces that text with "XY is a complete asshole." Now you probably didn't change your mind, but in Ted Nelson's scheme your web page would now suddenly read "Random Person says: "XY is a complete asshole." I'm inclined to agree."
Not really. As the rest of the world moves to IPv6, the US will eventually have to follow even while having IPv4 addresses available, or it will lose accessibility to a growing part of the net. And the rest of the world will have a head start on IPv6, due to being forced to.
In my view, the ideal interface would use three mouse buttons:
Left mouse button: Select. This includes selecting menu items etc. ...)
Middle mouse button: Modify (move, rotate,
Right mouse button: Context menu.
Of course there's the problem that with scroll wheel mouses, the middle button is combined with the scroll wheel, which makes that interface harder to use than with an old non-scroll three-button mouse. And of course, laptops almost always lack a middle button.
Do they? I would say, the learning curve describes the learning effort which has to be invested over the mastery to be achieved. A steep learning curve means a lot of learning for a little achievement. A barely noticeable incline would mean you are able to achieve great things with only a little learning.
Learning over time doesn't tell you anything about how difficult something is; it only tells you how hard-working or lazy the learner is.
The trick is to make a program usable without reducing its power.
Of course, adding keyboard shortcuts to an UI is trivial without otherwise changing the UI. Therefore simply add "with keyboard shortcuts added" to the OPs statement to deal with that.
I admit I have no idea what a subsurf preview is. However, is the limitation you mention really a limitation of the UI, instead of a limitation of the functionality? That is, would it be possible to add that functionality without changing the UI (apart from possibly adding another option somewhere)?
I have no idea what the Wings3D UI looks like, therefore I cannot say if it is really that good, but it seems to me that you didn't mention anything which would not be easily added when copying the UI.
I've never failed to find an unused spot (and some programs also hide the mouse pointer when working with the keyboard). Most programs have white space in the menu bar, many programs have scroll bars where you can put them, etc.
Moreover, I'm using "lazy focus-follows-mouse" where a window doesn't lose focus just because the mouse leaves it, but only when it enters another window. When I have non-maximized windows, I almost always have some unused desktop space around it where I can move the mouse cursor.
But then, most of the time I don't even bother to move the mouse cursor away completely. For example, as I write this, it's still in the input field. Only if it happens to disturb me at the place where it currently is, I move it away.
The main problem I have with non-mouse based focus is that the focused window always has to be on the top. I frequently put helper programs on top of the main program, at places where they don't disturb, and I hate it if they go away (hidden by the main program) every time I work with the main program. Now logically it would be possible to decouple keyboard focus and raised window even for click to focus, but nobody does (probably for the simple reason that those who dislike raise on click are those who prefer focus-follows-mouse anyway, so there would be no users using that feature).
No, MDI is (poorly) emulating a window manager. Unlike with MDI, on a WM workspace I can group windows from several different programs I use together. With MDI, this is not possible.
I don't have a problem with a program offering MDI for those poor people who cannot use a real window manager. However I have an issue with programs forcing an MDI on me, when my window manager is orders of magnitude better than the emulation the MDI program forces on me (partially by the mere fact that the MDI by design is limited to one application, and partially by the fact that I put considerable effort in the window manager to behave the way I like it, and of course the MDI application will ignore my window manager's settings, and that's of course ultimately also inherent, because no one can expect an application to know how emulate arbitrary window managers).
From other comments, it seems that GIMP did it the right way (let's just hope that they will never remove the "antiquated" independent-window interface for the sake of simplifying their code).
Silverlight doesn't work on Linux. Which means they can add features like crazy, I'm still not interested.
You say "a equals b" when you talk about assignment!? I definitely don't.
But a solution where imperfect humans are less likely to make errors to begin with is better than one where they are likely to make errors. Of course, the compiler should catch the error in both cases.
That's funny. I've gotten lots of mod points lately, definitely more than in earlier times. Maybe it's related to me using all of them. Because otherwise, the only thing I do is post comments (I didn't metamoderate a single post for quite some time, and after a few failed attempts I've given up on trying to post articles).
What about chatting with your girlfriend and pretending it's a bot? :-)
C# has UNSIGNED integers.
With sane semantics? Or did they simply copy the semantics from C, with all its drawbacks?
Did you warn them?
But a mistake you don't make is always better than a mistake that the compiler catches.
Not really. Object variables are references, and those references are passed by value. It may not be the way you normally think about it, but it is technically correct. You never pass objects in Java, neither per value or per reference. Of course, passing a reference by value is equivalent to passing the referenced object by reference, and generally you'll think about it in exactly that way, but that doesn't make his statement wrong.
That should of course have been "Up to here" ...
Up to hear I'm with you (well, not completely; instead of killing interfaces, they could be turned into something useful).
A definite NO! If there's anything bad about C/C++, it's the preprocessor.
Nothing against macros, but please not C/C++ style. And C/C++-style #include isn't a good idea at all; the only reason there's not yet a sane alternative in the C++ standard is that it wasn't considered ready.