Hi Xueqing, If I haven't been clear before this, thank you for your interest in this project. The non-linear writing enhancements have been a passion project of mine for quite some time. I am excited that there is another person interested in them and look forward to working with you toward their completion.
In this email, I am going to try and lay out their current status, next steps, and offer recommended components that might make a good proposal. To that end, I suggest you look at Google's guidelines for what the proposal should include: http://www.google-melange.com/document/show/gsoc_program/google/gsoc2013/help_page#5._What_should_a_student_proposal_look). Also, I think it a good idea to identify a few aims and specific features that you intend to work on through the summer. Completing and polishing all of the non-linear enhancements is quite a bit of work, and may require more than the few months available. I know that all of the LyX developers are hopeful that the summer of code students will become long-term contributors. I certainly am. # Project Overview If you haven't already done so, I would strongly recommend that you read through the non-linear writing essay at the link below: http://blog.oak-tree.us/index.php/2009/03/04/perfect-tool It provides a thorough overview of what I was hoping to accomplish with the non-linear enhancements (which I will refer to as LyX-Outline from here on out). From here, I will assume that you've read that original essay and are familiar with the general aims of the project. Since that original post, I've done quite a bit of work to move that vision forward. This includes an initial prototype (written in Python): http://blog.oak-tree.us/index.php/2009/05/13/lyx-outline # Current Status # An initial implementation of the core C++ classes needed to integrate the enhancements into LyX proper. Here is what has been completed: 1. There is a candidate implementation of an *integrated* corkboard and outliner pane that are connected to a single instance of a QStandardItemModel (http://qt-project.org/doc/qt-4.8/qstandarditemmodel.html). 2. To help facilitate communication between the view classes (corkboard and outliner), I've written a DataManager class that can be used to communicate with the LyX backend, which uses a message/action system to process events. All of these components work as expected (though there are some stability bugs which need to be addressed), and you can try them by building the program from the LyX-Outline branch on Launchpad: https://launchpad.net/lyx-outline Instructions on how to do this can be found at: http://blog.oak-tree.us/index.php/2010/06/25/lyx-outline02-1 This means that the *core* vision of what I described above is mostly complete. There are still a pretty large number of unfinished things, which are detailed here: http://blog.oak-tree.us/index.php/2010/07/19/lyx-outline02-3 But as a whole, they are pretty minor. Which leads me to a question of my own. I'm not sure I understand what you are asking in the section of your email below? > 2. About the "combining outliner and corkboard idea": I wasn't sure > what did that mean when I saw in the project description that the student > is expected to "integrate" the two methods. I'm glad that we are thinking > of the same thing and it is a very exciting idea. > In most ways, the outliner and corkboard are already integrated. They use a common model and get their data from the underlying document structure. What do you mean when you talk about "integration"? With that said, it would be nice if they were used a slightly better method to communicate with LyX's core. Right now, the model where they draw their data from is regenerated each time there is a major update to the document (such as the enter key or backspace key being pressed). This is done by scanning the in-memory representation of the doc and picking out certain portions (called insets). For larger documents, this is pretty inefficient and can lead to noticeable lags. A much better system would be to use signals and slots: http://qt-project.org/doc/qt-4.8/signalsandslots.html This would allow for various items in the model to subscribe to objects in the document and be notified of changes. At the time I did the original implementation of LyX-Outline, I decided not to take this route, though, because signals/slots were not allowed in the core of LyX. That has since changed: http://www.mail-archive.com/lyx-devel@lists.lyx.org/msg172533.html Modifying the LyX model so that a few insets, such as the text inset and note inset (which form the foundation of the document model used in the corkboard/outliner), would be a very good start to a GSOC project. It is nicely bounded with specific aims, and would provide a springboard to more ambitious goals, such as the draft interface you describe below. > > 3. About that "draft" interface to allow users to drag and rearrange > elements of the text, I haven't seen any application similar to my > idea. Maybe the closest idea to dragging blocks of sentences is how > you drag files in the list (either on a Mac or a PC) and rearrange > them, and in terms of dragging single words or phrases, think of how > you drag bookmarks on a bookmark bar in Safari. I personally would be > totally impressed if my text editor could do that. I think this sounds like a very nice enhancement to LyX. As of right now, the outline branch has a limited version of this feature, based upon defined sections (similar to the way the outline view in Word works). Each section may then have a note associated with it, and it's parent text. The notes, title, and text are kept in a model, and can be manipulated: re-ordered through drag and drop, indented/unindented, etc. If I understand your thinking correctly, you would like to extend this behavior to work on arbitrary snippets of text? Is that correct? There is a writing program called Scrivener which allows for a type of this where text snippets do not need to be tied to specific sections. See the link below for an overview of how it works: http://literatureandlatte.com/scrivener.php Is this similar to what you were thinking? If not, could you expand upon your original idea? How did you imagine splitting the writing into different chunks? What meta-data did you imagine being associated with each component? How do you imagine this information being rendered on screen? Would it follow similar metaphors as used by Scrivener, or a more abstract series of metaphors used by another writing program (such as Information Architects writer: http://www.iawriter.com/ipad/). # Next Steps # I understand that I've covered quite a bit of ground in this email and provided a deluge of links. Here are a few next steps that I think will get you closer to a project proposal. 1. You need to select a one or two *well defined* aspects of the non-linear writing enhancements that sounds of interest. One example might be to move the LyX core inset classes to QObject and adjust the TOC model to work on the basis of signals and slots. or, you might adjust the corkboard view to use QGraphicsScene: http://qt-project.org/doc/qt-4.8/qgraphicsscene.html The QGraphicsScene classes provide for a more performant base than QWidget (which we use now), and would allow for more flexible notecard views. Indeed, it might even be possible to create specialized views for more than just sections of text; figures, for example, might have one type of notecard (including images); and references, a third. Or you might decide to pursue something else entirely, such as what your draft view would require. 2. Once you settle on one or two major goals, we can start to talk about the best way to implement them, pick milestones, and discuss a timeline for deliverables. The more concrete you can be about the above, the better chance your proposal will have of being accepted. (See http://www.di.ens.fr/~baghdadi/TXT_blog/5_advices_to_get_your_proposal_accepted.lyx.html which has several pieces of advice for writing a strong proposal.) # Conclusion # As I said before, I'm looking forward to working with you and sincerely hope that your project will be approved. Discussing the non-linear enhancements has gotten me excited to pick up on several of my own LyX projects and to produce a killer writing environment. Cheers, Rob Oakes
