On 7/18/07, Bryce Harrington <bryce@...961...> wrote:
On Wed, Jul 18, 2007 at 10:33:00AM -0700, Ted Gould wrote:
Hello all,
As you guys are aware Chris is working on putting some bitmap editing tools into Inkscape for is GSoC project. There's a couple things that are difficult on the user interaction front that I wanted to bring up and see what people think.
Basically, there are two types of images: embedded and linked. There is two types of linked images that are read-only and ones that we can write. So what should happen when someone does a bitmap operation on these types of image? How should undo/redo be handled?
I'm leaning towards saying "always embed" at this point. It's simpler and probably does something similar to what most people want anyway. We don't need file dialogs or temporary files. What does everyone think about this? Chris, did I state the problem correctly?
Hi Chris,
Can you tell us more about your plans with this feature, and how embedding vs. linked will make things more challenging or less? Sounds like most people prefer embedded - does this lead us into any issues on your end we should know about?
Bryce
Embedding all images would certainly make it much simpler and easier. I don't know about speed issues -- but intuition tells me that having a 10 mb file would load more slowly than a 100 kb svg file and five 2 mb images. But that's a matter for someone who's put more time into Inkscape's rendering functions to answer.
If I inserted an image and executed some special effect on it, and then found out my program had converted my image into base64 and embedded it into my svg document, without asking me, I wouldn't be too I happy, personally. I don't want to make that the only option, and I'd actually rather not have it default.
My next goal is to be able to undo operations on linked images. If that isn't extraordinarily hard (and I don't expect it to be, I just have to get into the undo buffer, I guess), I don't see any reason why keeping images linked would be a bad idea.
-Christopher