| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Broken because of changed server return struct, now uses link_rm
dictionary (i.e. topo diff).
|
|
Broken during recent overzealous precondition addition
|
|
Doesn't fix node color, that is fixed in next commit
|
|
|
|
Still have separate paths for and links in parsing and to server. But a
one liner..
|
|
Also fixes name when none set (i.e. development mode).
|
|
|
|
|
|
pretty large commit, but it mainly boils to
- reusing the same format clone returns to the client for topo_diff
bad bad code introduced mainly in client side, but server wasn't spared:
- graph/commit_topo__diff massages the links too much - this should not
be there at all, it should be in the correct format to begin with.
server code duplication between clone & topo_diff post processing,
should be factored out
on the good side: single transaction whenever user presses enter, and
the correct place to send back any accumulated diffs (other user
changes).
|
|
|
|
|
|
|
|
|
|
|
|
if at all, not a confirm dialog)
|
|
|
|
|
|
cross the wire
|
|
|
|
Code is ugly on too many counts, but mainly:
1. repeats local change twice, once for temp and once for non-temp that
goes through server.
2. handles topo_diff reply expecting a single id instead of having a
general topo_diff reply handler that will take care of it, possibly
hanging a callback on it for graph update.
|
|
|
|
copied from update_node)
Adds a copied TODO for handling id changes (i.e. same problem on nodes)
plus a new TODO for a fake API (relevant to both nodes and links)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
directives for src names:
any js file should be relative
any resource should be absolute
i.e. all images are /res/img/bla
and all js files are bla.js / view/node_view.js etc
|
