GitLab

for garrays we disable editing gl_x1 and gl_x2. They get set automatically to show the entire array by default. May want to revisit array viewport later, but for now this keeps the bug from happening.

allow this to work:
gui_start_vmess("foo", "");
gui_start_array();
gui_s("hello");
gui_end_array();
gui_end_vmess();

In other words, a gui message with a single array argument. There should rarely be a need for this (if at all), but it should still send a well-formed message to the GUI (which it wasn't before this bug fix)

* canvas dialog is a bit too complex to have "Cancel" button revert any changes we may have applied. (It's not out of the realm of possibility, but for now it's easier just to use a nonstandard "Close" button to show the discrepancy)
* font dialog updates on change of radio button, so we just need a "Close" button there
* prefs dialog also too complex to do a revert on "Cancel"
* iemguis and gatom properly revert on "Cancel"

send old attrs before new attrs for gatom when clicking "Ok" on the dialog. This way the undo point gets set correctly even if the user has clicked "apply" beforehand

* still need to test with each iemgui
* still need to add flag for params callback to choose whether to have an undo_point

First stab at making the "Cancel" button in gatom dialog actually revert to the previously saved values.

* make scaling follow the mouse more consistently

fix jwilkes/purr-data#6 by listening for a "paste" event inside boxes being edited, and fetching only text from the clipboard
(need to investigate to make sure it works for drag/drop as well)

* simplify GUI API
* get rid of canvas_createobj

first attempt at lifting the limitation on the size of text typed in new objects in the GUI which get sent to Pd. We're sending in chunks smaller than the buffer to achieve this.
* this needs a lot more testing
* still subject to Pd's limit of 1000 bytes per t_symbol
* probably want to protect the new canvas methods from segfaults in case the users start fooling around with them

only put a new message box in floating mode (i.e., following the mouse) if it doesn't have an autopatch connection to the object above it

changed the interface for canvas_objtext, but it's not public so that shouldn't be a problem

changed the interface for canvas_objtext, but it's not public so that shouldn't be a problem

fixed with simple css change... definitely needs further testing to make sure there aren't any side-effects

* don't put patch body in index... just use keywords
* index entire "doc" directory (recursively)

take patch file content out of the search index, clean up dialogwin references when window gets closed

first draft of a search engine (powered by elasticlunr). We're just using the release file for elasticlunr here. It might make more sense to do it as a gitsubmodule.