umlsynco / umlsync-framework Goto Github PK

• Add support of tree remove
• commit tree modification (without single item)
• Add checker for files existence in a upper tree

• Module for Eclipse plug-in
• Dialog for connection
• Tree navigator
• Tree switcher (Project navigator, CDT, JDT)
• Content loader
• Think about database API usage (dialogs and controllers)

• another buttons for embedded diagrams
• another actions for embedded diagram

Single canvas support for not embedded content

Farmework has a number of content handlers therefore we should simplify view open, but with embedded flag on.

For the diagrams, we should avoid DND support for embedded content.
For the code snippets we should avoid edit

etc ...

• It looks very tricky when we use jQuery.dynatree for the backbone view. But for another hand it is really easy to re-write it with Marionette (Use model.cid for the corresponding content and parentCid fo embedded content)

• propagate add/remove/change events for application manager
• get JSON or SVG description of model

• Make a commit of modified files on github
• Reload tree

localhost + port + secret ?

Prototype operation manager support:

• sync-up model on D&D, resize and edit
• subscribe on events
• modifications queue (bi-direction)
• view modification on model-change

1. Pull the latest version of the repository/branch
2. Highlight all conflicted items
3. Identify the way to resolve conflicts

• Provide and interface for the access to the local clone of repository (tree and file)
• Provide cscope cmd line based interface

The current implementation of the content cache collection suppose only one instance of each file, but in real case we could have several blobs-sha for each instance. So, the major idea is to use sha+path as content key

Try to re-use snippets from umlsync:

• storage model
• dialog

Make an embedded diagrams support for markdown

• ll-conector DND should not take extra elements
• merge ll-ports
• extend llport on connector drop nearby
• fix the connector's menu dnd coordinates
• cross llport DND in case of icon menu or not ?

• Add Edit/view mode support for the diagrams
• Make view only mode for embedded diagrams

Open an existing content and close it

Add context menus for the connectors:

1. Remove
2. Add Label

• Implement ref-counters for the content cache models
• implement time stamps for the content cache models
• implement cache garbage collection strategy

• extend context menus description asset
• provide "diagram:connector-label:[remove|edit]" handler
• extend context for each label (provide diagram & connector)
• make vent available on the LabelView level (there is no way to take framework vent directly )
• connector highlight on mouse over the label

• google codesearch project provide cindex/csearch cmd line utilites
• snavigator has hyper cmd line too

It is necessary to implement API for the indexing requests and handle them via cmdline wrapper

Embedded content frame size handling

1. New diagram
2. Save/close
3. Open
4. Try to add field or method

• remove blob/tree support
• Modify update procedure
• Modify post-processing procedure

Add the context menu controller for the GitHub tree

• - getUmlLabels - as collection of { name, x, y}
• - label position on load/save/restore
• - label DND support (including model {x, y} change )
• - label editable ( model change "name")
• - connector highlight on mouse over the label
• - DND label on connector DND (multiple DND use-case only)

Port an UML creation accordion-menu

There is no status change for content on modification

• Add content to the tree
• Commit newely added content

• Handle create new content controller

• on edit/view mode
• on tab change
• for markdown and code views

• mouse over the connector label and connector line itself
• start DND
• alert happen

• Extend the load content controller to save new content
• Create Dialog for content selection
• Create controller for the content creation/open
• Path load helper controller to get information about sub-paths

• Add support of content handling in the file-tree view

• Add support of modified/updated, new, removed and cached status

• color change
• alignment change
• Font change
• Z-index change

• resize left-righ
• resize window
• open developer console F12
  etc ...

Commit on the GitHub consists of several steps:

1. Pull latest changes and fix conflicts (another controller)
2. commit all blobs (content cache responsibility)
3. modify and commit tree
4. make a commit instance
5. PUT commit on the HEAD of the current branch

The current content loader mechanism doesn't support embedded content load:

1. There is no command to load
2. There is no cached content reference counter

The major idea was to split DataProvider(DT) and ContentTabs(CT) viewer.
It is some kind of RPC between DP and CT. And we need to define a new on for the embedded content.

1. Obj instance
2. Alt, Cond, etc
3. llport, del, lost/found message
4. connectors

Question: How to handle obj instance creation ?

Sometimes it is possible to get double instances of tree items in a single view.

Open tree vi JSONP requests:

• Models to get JSONP requests
• Configure TreeView

• wrong path in markdown
• failed to load content data or path
• wrong path (doesn't exist)

Icon menu connection helper has wrong position

• Backbone model of the diagram, elements and connectors
• Views for elements
• view for connectors
• Behaviors
• Module based loader (partially destroy)
• configurable behaviors

Add the content "GetLink" button support to cache path to content

1. It should provide the way to copy data manually
2. It should cache data internally too (To press "D" button of the markdown menu)

In the future "GetLink" should point on the content path which I could send to colleagues or post them on external resource

• Create the context menu for elements
• Add extensions for menu if diagram belong to some specific source code provider (Extend with indexing)