0263 (3.3.6 or 3.4)


contrib
X shell discussion with gohai
X   https://github.com/processing/processing/pull/5082


_ add separator option to loadTable() 
_   https://github.com/processing/processing/issues/5068

_ add blendMode() to nope() calls in PGraphicsPDF
_   https://github.com/processing/processing/issues/5105

_ WARNING: GL pipe is running in software mode (Renderer ID=0x1020400)
_   is this coming from us? if so, need to provide actions


high
_ implement sketch scaling into PApplet
_   https://github.com/processing/processing/issues/4897
_ Sketches on Windows don't take UI sizing into account
_   https://github.com/processing/processing/issues/4894
_ Sketches on Linux don't take UI scaling into account
_   https://github.com/processing/processing/issues/4895

_ TRIANGLE_STRIP not working correctly with createShape() and default renderer
_   https://github.com/processing/processing/issues/4678

_ should we drop the 'default' prefix from the ex handler so it's not static?
_   http://docs.oracle.com/javase/1.5.0/docs/api/java/lang/Thread.html

_ option to disable OpenGL ES on Linux? 
_   https://github.com/processing/processing/issues/4584

_ textAlign(CENTER) and pixelDensity(2) aligning incorrectly with Java2D
_   https://github.com/processing/processing/issues/4020
_   calling textSize() fixes it

_ should fullScreen() set width and height to displayWidth/Height
_   or is that being set/unset used for any state info?
_ when calling exit(), if sketch has halted w/ exception, force the quit
_   otherwise have to double-quit with cmd-q on OS X
_   simple test case: using size() inside setup() from Eclipse


discussion
_ how to handle the touch api
_ figure our size(img.width, img.height) situation
X   just make loadImage() work in settings
_   update the wiki and reference
_ size() function that scales to screen, keeps aspect, re-scales mouse coords


known
_ window must close when using file dialogs with OpenGL on Windows
_   https://github.com/processing/processing/issues/3831
_ window loses focus after maximizing
_   https://github.com/processing/processing/issues/3339


misc
_ move blending calculations from PImage into PGraphics
_   tricky because that means moving blend_resize() as well
_   and should that live in PGraphics or be its own class or ??
_ try using Frame instead of JFrame
_   default sketch location using insets incorrectly
_   use the BufferStrategy directly from the Frame object?
_     might fix performance issues w/ Presentation mode
_ sketch placement (window insets) not properly set on Linux
_   https://github.com/processing/processing/issues/2063
_ note in docs that full screen and present are now different
_   on Export to Application, this has an impact
_ update wiki/docs to say "don't override sketchXxxx() methods"
_ SVG only exports last frame
_   possibly because Java2D is disposing the Graphics2D in between?
_   https://github.com/processing/processing/issues/3753


javafx
_ wrong window size with fullScreen()
_   https://github.com/processing/processing/issues/4737
_ menu bar not hiding properly in exported applications with FX2D
_   https://github.com/processing/processing/issues/4527
_   hideMenuBar() called from setup() works fine
_   just call it around setup time? 
_ the --hide-stop option not working (FX only? traces?)
_ make wiki about quirks
_   starving the thread makes things really slow down
_   keyPressed() is always uppercase, keyTyped() will be correct
_ do we really need setTextFont/Size when we already have Impl? 
_ need keyPressed() to do lower and upper case
_ static mode sketches (draw once and halt w/o closing window)
_ fix display handling, line up the device order with AWT
_   https://docs.oracle.com/javafx/2/api/javafx/stage/Screen.html
_ noLoop()
_ present mode not working at all
_   stage in the center, clear the rest of the screen
_ createGraphics() should probably create PGraphicsJava2D
_   or is Canvas specific to the PGraphics, and we get another Context2D?
_   http://docs.oracle.com/javafx/2/api/javafx/scene/canvas/Canvas.html
_ loadPixels() (also 2x)
_ text and fonts?
_   maybe helpful: https://wiki.openjdk.java.net/display/OpenJFX/Font+Setup
_ updatePixels()
_ save() and saveFrame()
_ get() and set()
_ clip/noClip
_   https://github.com/processing/processing/issues/3274
_ getNative() in PImage problematic because it gives back a BufferedImage
_   move loadImage() into PGraphics, with AWT version the default
_   or pass createImage() through to renderer?
_ implement external messages (moving the window)
_ implement PSurfaceFX.setIcon()
_ javafx not supported with ARM (so we're screwed on raspberry pi)
_   https://www.linkedin.com/pulse/oracle-just-removed-javafx-support-arm-jan-snelders


opengl questions
_ exitCalled() and exitActual made public by Andres, breaks Python
_   also not API we want to expose, so sort this out
_   or maybe we're fine b/c now FX2D needs it as well
_ when did setPath() sneak into PShape? API is nothing like anything else
_   probably from the material stuff, but we need to fix that
_ why is createShape() implemented 4x (for P2D, P3D, and 2x versions)? 
_   shouldn't be static, run it from the renderer, that's point of createXxx()
_ public createShape() method that takes another shape as param?
_   should just be the constructor doing this, or copy()


full screen
_ present window draws in stages (OS X)
_ crash on startup when "Mirror Displays" selected (cantfix?)
_   suspect that this is a specific chipset since Oracle didn't reproduce
_   AMD Radeon HD 6770M was in the Oracle bug report
_   https://github.com/processing/processing/issues/2186
_   https://bugs.openjdk.java.net/browse/JDK-8027391
_   test with JG's 13" retina laptop


graphics
_ need to be able to call pixelDensity(2) before beginDraw()
_   add pixelDensity() method to PImage/PGraphics
_   pixelDensity(2) does: pixelWidth = width; width /= 2; pixelDensity = 2;
_   this works for both PGraphics and images (though they're a little backwards)
_   for PGraphics it comes early enough
_ should the re-alloc of the drawing surface happen in beginDraw()
_   that way it won't blit to the screen until we have a fresh redraw?
_   otherwise it'll also be resizing on another thread.. badness
_   how to allocation image object w/ createGraphics() (since no surface)
_ createGraphics() currently broken in Java2D
_ size() command not working to do a resize
_   need a programmatic way to set size
_ deal with applySettings() change (maybe not a problem?)
_ check on performance of the new EDT setup
_ present mode is 30-40% slower than windowed
_   w/ this example: https://github.com/processing/processing/issues/2423


data
_ implement strip(), lstrip(), rstrip?
_ handling of 'missing' values in Table needs serious work
_   if missing int is zero, can't just remove those values from saving a table
_   but for NaN values, it's a necessity
_ get() methods in List/Dict shouldn't allow you to get bad values
_   but set() methods can automatically resize the arrays
_   though that means insert() should allow you to insert past the end
_   and should add/div/mult let you work on non-existent elements?
_ addRow() is not efficient, probably need to do the doubling
o   or have a setIncrement() function?
_   it would default to 1 on tables loaded from a file
_   and default to doubling when created with "new Table"
_   row count and array size are combined.. need to break apart
_ match and iterators
_   add match version that returns table that's only a pointer to original
_   save the constructor for the version that actually copies data
_   the table pointer version will be speedy and allow chaining 


high
_ point() rendering differently in 2.0.3 and 2.1
_   https://github.com/processing/processing/issues/2278
_   internally, we probably have to call set() if it's a 1 pixel point
_     but that's going to be a mess.. need to first check the CTM
_ tint() not working in PDF (regression between 2.0.3 and 2.1)
_   https://github.com/processing/processing/issues/2428
_ finish PFont.getShape() implementation
_   needs to have a way to set width/height properly
_   draw(s) doesn't work on the returned PShape
_ y-location of frame might be negative, but that might be ok 
_   right now, on Windows it moves it on-screen (b/c of previous bug reports)
_   may be cause of some of the display placement issues w/ multiple displays
_   seem to recall one of the bugs mentioning stacked displays
_   need to change to iterate through display rectangles
_   may be related to full-screen bug filed recently
_ closing a sketch window manually (not hitting Stop) not killing sketch
_   at least with the Java2D renderer
_ don't override the window icon w/ p5 logo if already set


decisions/misc
_ Separately, if we wanted, we could add a method that can safely do drawing calls, but that won't be any faster or make use of more processors the way that `thread()` does. 
_ need reference update for createShape()
_   though we need to verify the behavior here
_   createShape(RECT) uses x/y/w/h, and optionally adds a mode param
_     and optionally the corner radii
_   createShape(ELLIPSE) is x/y/w/h, and optionally adds a mode param
_ async requests
_   https://github.com/processing/processing/issues/3165
_   Request r = createRequest("http://p5.org/feed/13134.jpg");
_   Request r = createRequest("http://p5.org/feed/13134.jpg", "callbackName");
_   Request r = createRequest("http://p5.org/feed/13134, "callbackName", "jpg");
_   implement callbacks for images and other loadXxxx() functions
_     remove requestImage() and friends
_   callback for requestImage()
_     https://github.com/processing/processing/issues/680
_   remove/update requestImage example
_   check into promise api
_     http://www.html5rocks.com/en/tutorials/es6/promises/
_     https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Promise
_ fix the registered methods doc, stop/dispose working
_   Casey needs to nudge people about libraries, so we need to fix this
_ pause(), resume(), start(), stop() clarifications
_ dispose/stop not great w/ libraries yet
_ PShape complete refactoring
_ proper touch events
_   touch event doesn't make sense for mouseMove on desktop, hover on Android
_   probably go with pointer: more universal for touch/mouse
_ high-level touch events (swipe, et al)
_   if Andres is done, they go in, if not then 3.0
_ move away from loadPixels
_ add options for image.save() (or saveImage?)
_   add quality=[0,1] for jpeg images
_   add dpi=[0,n] for for png images
_   implemented internally, just needs the wrapper for the args
_ should map() actually constrain to the low and high values?
_   or have an alternate version that does that? (or boolean param at end?)
_ decide whether to keep:
_   public float textWidth(char[] chars, int start, int length)
_ add version of math functions that use doubles?
_   what other methods should work with doubles? all math functions?
_   seems like internal (mostly static) things, but not graphics api
_ new PGraphics(... OutputStream) 
_   https://github.com/processing/processing/issues/285
_   already added for PDF, just need to work out the API
_ if save() returns boolean, does saveFrame()?
_   also need to copy this over to android
_ "translate, or this variation of it" when using text(s, x, y, z) accidentally
_   change to be the text command
_ size() and resize() and whatever? 
_   should be setSize(), but that's odd for image files
_   -> resize() is fine with PImage and PGraphics... 
_   we may be inheriting the resize() from Java -> make it final?
_   add resize().. make it call setSize().
_     also needs to do a redraw if noLoop() has been called
_   the registered method for size() also needs to be called
_ get() or copy()? for vectors, image, etc.
_ toArray(), toArray(float[]), toVectorArray(), toVectorArray(PVector[])
_   toColorArray(), toColorArray(float[])...
_ make sure that loadXxxx() methods are used after init()
_   nasty errors when loadImage/Font/createFont/etc used outside
_   decision: add error messages where possible
_   idea: set frameCount to -1 when setup not run yet?
_     then set frameCount to 0 when setup() starts?
_ need to clean up the hints in the reference/source
_ exactly how should pixel filling work with single pixel strokes?
_   http://dev.processing.org/bugs/show_bug.cgi?id=1025 (no gcode)
_ y2 position of rectangles not same as y2 position of lines
_   happens when the rectangle is flipped on the x or y axis
_   probably a hack that draws the "last" point differently
_ add inputPath() and outputPath() -> sketch folder or sd card
_   or should this just be a change to sketchPath() on Android?
_   also because input/output won't be different (since not data folder)
_   or should this be storagePath() etc?
_ PUtil -> move match(), lots of other non-gui functions into own class
_   and merge it in statically via the methods code
_ check for what else inside PApplet should be static
_   maybe catch RuntimeExceptions in the called sub-functions
_   that way more stuff can be static w/o losing useful error handling
_   (emitting errors when closest to source.. i.e. w/ the filename)
_ change how beginRecord() works.. passing around PApplet vs PGraphics is gross
_   have to pass PApplet just to make the rendering work to both renderers
_   should instead be a renderer that splits things out
_ wrap exceptions with die() and warn() methods
_   this way, people can make use of exceptions if they would rather
_   or shut them off if they don't want to 
_   also need to explain exception handling in general
_   https://github.com/processing/processing/issues/222
_ bring PConstants back in line w/ previous 1.5 (can't renumber)
_ finish adding loadStrings(BufferedReader)
_   decide if we want to keep this (and/or others?)
_   port to android
_ deprecate the older splice() etc API inside PApplet?
_ add notes about the new shader setup
_ add reference/docs for urlEncode() and urlDecode()
_ add explanation to the reference about using beginRecord() with fonts
_   pdf.textMode(SHAPE)
_   also set the font *after* the record has started
_   maybe should instead make textMode(SHAPE) the norm? 
_   and people can change it to textMode(MODEL) if they want?
_   http://dev.processing.org/bugs/show_bug.cgi?id=1535 (no gcode)
_ explain the new PGL interface
_ decide how disconnectEvent should actually be handled (and name?)
_   was disconnect always there? 
_   will need documentation
_ negative indices so that we can work relative to the end in data classes?
_ add Double and Long versions of the data classes?
_ add requestFocus() method to PApplet (or surface?)



////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////



CORE / rendering/performance/threading

_ hint(DISABLE_LOAD_PIXELS) -> faster rendering in Java2D
_ problems with 2D rendering speed
_   volatile images
_   http://www.javalobby.org/forums/thread.jspa?threadID=16840&tstart=0
_   http://www.javalobby.org/forums/thread.jspa?threadID=16867&tstart=0
_   http://www.gamedev.net/page/resources/_/technical/general-programming/java-games-active-rendering-r2418


CORE / stop() 

_ in PApplet.main(), windowClosing() should probably be calling 'exit()'
_   or rather, we should never call System.exit(0), ja?
_ dispose() method in PApplet should be empty so ppl can override
_   move that stuff to destroy()
_ pause()/resume() need to work on the desktop side as well
_   notify ddf when pause/resume implemented
_ stop() not called in 1.5
_   https://github.com/processing/processing/issues/675
_   noticed that dispose() is getting called. stop() isn't.
_   http://java.sun.com/docs/books/tutorial/deployment/applet/lifeCycle.html
_ static mode sketches seem to break ESC... noLoop() problem?
_   need to find another way to get ESC on static mode
_   b/c static mode sketches *do* finish because they have no draw()
_ sort out destroy(), stop(), pause() et al
_   ColorSelector should stop/pause when not visible
_     right now it's doing a low-level looping
_   start()/stop() perform like onPause()/onResume()
_     all of which call pause() and resume()
_   destroy() (from Applet) calls (our) dispose()
_ destroy() shouldn't call exit()... change from lonnen
_   calls ((PApplet)this).exit() instead of stop() (since stop is pause)
_ notes
_   exit() should do the actual exit
_   if inside draw, let it finish the loop
_   if not looping, need to do it immediately
_   does stop() unwind the thread, or does the thread unwind call stop?
_   browser will call start() and stop() methods
_ need to figure out start/stop signals coming from the browser
_   is it dispose/destroy? 
_ when closing a sketch via the close box, make sure stop() getting called
_   test to see if it's working
_ what's up with stop() vs exit()?
_   need to get this straightened for p5 (i.e. bc has this problem)
_   make sure the main() doesn't exit until the applet has finished
_   i.e. problem with main calling itself multiple times in Alpheus
_   if exit() (or stop) is called, then System.exit() gets called,
_   even though the main() wants to keep going
_ hitting ESC in a running noLoop()ed sketch won't close the sketch?
_ noloop ref even says that redraw will be called on resize, make sure it is
_ focus not coming through, ESC no longer working(?)
_ hitting cmd-q when an applet is running quits p5 (on macosx)
_   but cmd-q when running externally is ok because it just quits
_   is there a way to catch cmd-q when running a sketch? 
_     so that it could avoid quitting if the sketch hasn't been stopped
_     or if the sketch window is foremost
_     maybe a hack where a new menubar is added? 
o destroy() removed, but bring back? is that better than dispose()?
_   destroy() only called dispose(), so no difference
_   Python Mode has a hook for when it's called


CORE / PFont and text()

_ font rotation (native font problem?) with natives?
_   https://github.com/processing/processing/issues/731
_ Text rotation, placement and font metrics incorrect when scaled
_   https://github.com/processing/processing/issues/2167

_ remove subsetting stuff from PFont?
_   or use hint(ENABLE_FONT_SUBSETTING)? 
_ what's the difference with ascent on loadFont vs. createFont?
_ svg fonts
_   would be nifty if we could convert on the fly to ttf for speed...
_   https://github.com/processing/processing/issues/787
_ text position is quantized in JAVA2D
_   text placement is ugly, seems like fractional metrics problem
_   https://github.com/processing/processing/issues/138
_ text(char c) with char 0 and undefined should print nothing
_   perhaps also DEL or other nonprintables?
_   book example 25-03
_ accessors inside PFont need a lot of work
_ improve font metrics
_   http://java.sun.com/products/java-media/2D/reference/faqs/index.html#Q_How_do_I_obtain_font_metrics
_ font encoding issues 
_   java seems to force straight windows encoding.. (problem for pi fonts)
_   opentype/cff fonts don't work with live loading from the app
_   many (all?) opentype fonts won't show up or aren't supported
_     this may be only cff fonts that have trouble
_   when encoding is not the standard encoding, problematic
_     so sonata otf and sonata don't seem to have any chars at all
_ text() with a z coordinate is now using translate, very slow
X   also puts up a weird error message about translate() in 2D mode
_ make screen space fonts use get/set as well?
_   too much to debug on their own
_   unfortunately tint not set with setImpl, but...
_ not having kerning really blows
_   could this be pulled from the OpenType font stuff? 
_   it could be placed at the end of the file
_ not having fractional widths on small fonts really blows
_   screen space text looks crappy
_ working with vector fonts? 
_   need to be able to handle shapes within shapes (reverse winding)
_   ftgl: main code is in FTVectoriser
_   uses gluTessBeginContour and gluTessEndContour
_   and then does 5 step sizes for each curveto
_ show an error when using a font character that isn't available
_   maybe fall back on other characters instead?
_ some font weights cannot be specified/selected on OS X
_   https://github.com/processing/processing/issues/1727


CORE / PImage

_ TGA files writing strangely
_   https://github.com/processing/processing/issues/2096

_ loadPixels() implementation needs to be in PApplet, not PGraphics
_ this is a tricky thing to implement because of how OpenGL is handled

_ loadImage() should use the faster loading methods
_   hint(DISABLE_IMAGE_CACHING)
_   add a note to the loadImage() reference page
_ figure out why 1024x768 image takes 3.5 seconds to load
_   would using a BufferedImage work better?
_   is the image actually a BufferedImage so PixelGrabber is a waste?

_ deprecate the blend() function

_ implement unified means to specify smoothing level for image scaling
_   https://github.com/processing/processing/issues/204


CORE / PShape

_ major refactoring for PShape/PShapeOpenGL
_   https://github.com/processing/processing/issues/1623
_ PShape should be cached as well
_ major surgery to put loadShape() back into PApplet/PGraphics
_   then have the OpenGL or Java2D versions as cached objects
_ PShape s = createShape(); / s.beginShape(QUADS);
_ keeping these public for now
_   get/setNormal
_   get/setTextureUV
_   get/setFill
_   get/setStroke
_   -> perhaps these go into their own section?
_   -> or have their own accessor that returns all params?
_ trying to remember why primitive was changed to kind? (better name?)
_ these aren't per-vertex:
_   get/setStrokeWeight
_   get/setAmbient
_   get/setEmissive 
_   get/setShininess
_ boolean isGL() -> now removed
_ unapproved (made protected)
_   static shape copying methods
_   getTop, getBottom, etc
_   void solid(boolean)
_   void setParams()
_   void setPath()
_ colorCalc() methods added to PShape.. should just be used from PGraphics
_ loadShape() needs to live in PApplet
_ make PShapeOpenGL a cache object
_ don't allow things inside begin/endShape() that aren't possible
_   better lockout inside beginShape() to keep other things from happening
_   don't allow you to draw stroked items unless stroke() is called
_   don't allow beginShape() if shape is already set
_     (otherwise will cause some very strange errors)
_   https://github.com/processing/processing/issues/174
_ make determination on shape(x,y,z,w,h,d) or no
_ PShape API to handle internal vertex stuff
_   add deconstruct() method for paths
_   toArray() and toVector()
_   setVertex(0, x, y), setVertex(0, x, y, z)
_ if PShape3D, then is it PShape2D? or do you handle both the same?
_ add methods to PShape to apply all transformations in the tree
_ we can do hit testing (at least in 2D) now that we rely on java2d
_ for subclasses, make it easy to grab the structure of vertices
_   actually, just use PShape internally and this will be easier
_ for PShape, need to be able to set the origin (flash people)


CORE / PShapeSVG

_ implement support for SVG gradients from Inkscape
_   https://github.com/processing/processing/issues/1180
_ need to handle <!ENTITY tags in XML for SVG documents
_   get entity tags working in xml library for SVG
_ cover the svg changes in a future release
_ breaking up classes / figure out how to handle fonts subclass
_ when using get(), reset the bounds for the objects
_   otherwise it's always relative to the original document
_ support for text (shouldn't be too bad, use createFont)
_   implement text spans for svg output
_ add better support for attributes buried in styles (support ai9/10/11)
_ test what happens when transparency is used with gradient fill
_ look into transformation issues... guessing this is probably wrong
_   this may be what's throwing off the radial radius transform
_ check for any other pieces of missing path api
_   multiple sets of coordinates after a command not supported
_   i.e. M with several coords means moveto followed by many linetos
_   also curveto with multiple sets of points is ignored
_ document somehow.. svg viewer will be discontinued
_   http://www.adobe.com/svg/eol.html
_ PShape getVertex() not implemented properly for SVG files
_   https://github.com/processing/processing/issues/1596


CORE / PVector

_ add screen(PVector), model(PVector) and world(PVector)?
_   maybe screenVec()? or screenXYZ()?


CORE / OpenGL (Andres)

_ textureWrap(CLAMP / REPEAT)
_ simple NPE in OpenGL causes really large, not useful, stack trace
_ why is initPrimarySurface() public?
_ why is setFramerate() public? (also should be setFrameRate or just frameRate)
_ where did hintEnabled() come from?
_ remove 'params' from createImage (is it on loadImage too?)
_ remove matrixMode(), add a projection() method
_ shared intf for 3D view data across PGraphicsOpenGL and PGraphicsAndroid3D
_   libraries have to do a lot of casting
_ opengl isn't drawing rectangles out to raw properly with fonts
_   when not in textMode(SHAPE) should have rects
_ InvScreenX, Y, Z to convert screen (mouse) coordinates to model coordinates
_   https://github.com/processing/processing/issues/1609


CORE / Events

_ touch events.. can't do MouseEvent et al with Android
_   http://dvcs.w3.org/hg/webevents/raw-file/tip/touchevents.html
_   http://www.html5rocks.com/en/mobile/touch.html
_   decision: go with what looks like javascript/ios
_   touchEvent(), gestureEvent()?


LIBRARIES / PDF

_ beginRecord() doesn't use current settings of the sketch
_   sometimes reported as a bug, but probably not a good way to 
_   consistently carry these over
_ pdf sketches exiting before writing has finished
_   people have to call exit() (so that dispose() is called in particular)
_ when using noLoop() and the PDF renderer, sketch should exit gracefully
_   because isDisplayable() returns false, there's no coming back from noLoop


LIBRARIES / Video 

_ remove dispose() being used in the Movie and Capture
_   added warning notes, but shouldn't be necessary
_ buffer sink methods in Movie and Capture
_   added warning to JavaDoc, but need proper API for it
_ 'newFrame' is 'available', and ready() is part of that



////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////


DATA / Misc

_ add indent= as option for XML save
_ add indent= as option for JSON save
_ not doing print() methods, since alternatives are more descriptive
_   println(obj) and obj.write(System.out) are both better
_ can we use String... for options? or does it screw up the method signature?
_   and would we have to always concatenate arrays to prepend extensions, etc
_ include Instant and Interval? (or just Time and TimeSpan)
_ it's going to be File or Reader (mostly BufferedReader) everywhere
_   though TableODS needs an InputStream... 
_   and XML could use InputStream if we hope to be able to reflect doc encoding
_   ok, so File, Reader, and InputStream (where needed)
_ setMissingXxxx() -> should this live in PApplet? be static?
_   cons: static stinks, diff tables might use diff values
_ will also need an iterator for the Dict class ala Map.Entry
_ using Iterable for rows(), keys(), etc
_   generally more efficient for Table, but slower for FloatHash and IntHash
_   could use an int array instead, but a bit hokey in places
_ Iterable, Iterator, or [] returned for keys(), rows(), etc.
_   list, dict, json, table are each more efficient at different things
_   keys(), rows(), etc. should return something Iterable
_     whether that means an array or an actual Iterator or even a Set
_     that way we can do what's most efficient
_   then we add keyIterator() and others to handle the other case (deletion)
_   blogs.oracle.com/CoreJavaTechTips/entry/using_enhanced_for_loops_with
_ means that JSON.keys() -> JSON.keyIterator(), JSON.keySet() -> JSON.keys()
_ what should they all return? 
_   remove -> true/false based on whether it found something?
_   remove all -> the number removed?
_ List: remove(), append(), index(), etc all use values
_   removeIndex(index) is the other
_   otherwise remove() would be the only one that dealt with indices
_   Dict will use remove(key), so using remove(index) as default
_     and removeValue(value) is probably used less
_ introduce remap() instead of map() (if not deprecate?)
_ key/valueIterator to get a version that can be removed in a loop
_   in Table, added an Iterator function
_ should keys() returns cropped internal array after all?
_   right now it's making a copy (which is safer, so folks don't modify)
_   but this means that we don't have a fast method for access
_ methods for max, min, index of max/min entries? valueIndex?


DATA / Table

_ function that will convert awful CSV to TSV.. or something else?
_   maybe to write binary instead? then read the binary file once it's ok?
_ if loading from a File object
_   (or PApplet is passed in and we can check online)
_   then check the (probable) size of the file before loading
_ no column max/min functions since it needs to be per-datatype
_   better to use float mx = max(float(getColumn(3)));
_   but what to do with null entries?
_ need a method to reset the row/column indices after add/remove
_   or just make sure that it's covered for all cases
_ prefixes like # as comments? easy to remove?
_ naming for these (or whether to include hash)
_   Table createSubset() -> leave out?
_   String[] getUnique(col) (SQL DISTINCT, sort --unique)
_   HashMap<String,Integer> getUniqueCount(col)
_ really need to get an option to grab matches based on a col
_   i.e. get unique tables based on a particular column
_   or, get uniques, then grab rows that match a name in a particular col
_ create table from TableRow iterator...allows for subset and find
_   downside is types are not included
_ getMaxFloat() (whole table) or getMaxFloat(col) 
_   that's max(getFloatColumn(n))
_   also important b/c can leave out missing values
_ dictionary support
_ join tables together (static is kinda gross)
_   table.append(anotherTable)?
_   table.join(anotherTable)
_ getInt() on categorial to return index?
_   getCategories() and getCategory() methods to query names?



////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////


LATER / Optimize

_ put a wrapper around InputStream returned by createInput
_   that actually has some notion of the length of its input
_   that way, methods could use the information when reading
_ loadBytes() needs optimization
_   don't bother using a buffering stream, just handle internally. gah!
_ can loadBytes() be improved by querying file size first?
_   background 
_     this would require a new version of createInput(), which would query
_     the URL (or file) for an actual file size. the size is not always
_     available, so it can't be relied upon, but would help in some cases.
_   loadBytes() is used for images.. ouch
_   might be worth doing a test to see if it actually would help at all 
_     before rewriting all of createInput()


LATER (post 2.0)

_ fillMode(NONZERO) and fillMode(ODD) to replace solid(boolean)?
_ create PVec2D and PVec3D as separate classes
_ gui priorities
_   + Label
_   + Button
_   + Scrollbar/Slider
_   - Checkbox
_   - Radio
_   . List
_   . Textblock
_   \ Knob
_   \ Progress bar
_   \ Toggle
_ add shader support
_ createColor() instead of color()?
_ route all exceptions through PException and catch method
_   advanced users can override the method if they want
_   or you can set an option to have PExceptions be raised
_   decision: just copy & paste the serial/net code.. 
_     until we can find a more compelling example
_ actual shape api for things like rectangles and whatnot?
_ should we kill import xxxx syntax for libraries? 
_   just give up and use a gui for it
_   cons: copy & paste breaks
_   pro: there's no guaranteed 1:1 between packages and single libraries
_ method of threading but queue an event to be run when safe
_   e.g. queueing items like mouse/keybd, but generic fxns
_ for begin/endRecord, use a piggyback mechanism
_   that way won't have to pass a PApplet around
_   this has a big impact on the SVG library
_   in fact, this maybe should be a library that does it
_     so that the file size can be much smaller
_ add setOutput() method across other renderers?
_ introduce calc()
_   semantics of this are just too tricky, especially when it's not clear
_   what's actually gained by the split--would have to do weird hacks
_   to get the accum buffer, etc. to work anyway
_ add some sort of unprojectX/Y/Z method (based on glu fxn)
_   http://processing.org/discourse/yabb_beta/YaBB.cgi?board=OpenGL;action=display;num=1176483247
_ add a timer(obj, "functionname", 45) method
_   this can be used to schedule something to happen at a specific time
_   or import the swing timer (for actionPerformed)
_   also add interval("functionname", 40)
_ should fill(c) instead be fillColor(c)?
_ should color(123, 4, 99) instead be createColor()?
_ rounding errors on color conversion
_   colorMode(RGB, 1.0); colorMode(255); println(red(color(0.5,1,1)));
_   will return 127, instead of 128.
_ gray that's greater than the colorMode() can produce strange colors
_   http://dev.processing.org/bugs/show_bug.cgi?id=432 (no gcode)
_   http://processing.org/discourse/yabb/YaBB.cgi?board=Syntax;action=display;num=1083650609
_   http://processing.org/discourse/yabb/YaBB.cgi?board=Syntax;action=display;num=1082481891
_ add stroke() to type
_   more for doing outlines-only and that sort of business
_ gradients
_   use gradients in a class for the svg lib for now
_   don't add api to main p5 stuff (though maybe setPaint/noPaint api?)
_   gradient-painted lines and fills
_     java2d will do both line and fill, illusfarter only does fills
_   gradients not supported in java2d
_     http://dev.processing.org/bugs/show_bug.cgi?id=371 (no gcode)
_ bspline or nurbs (later, want to do the 3D/arch stuff correctly)
_ non-homogenous coloring for curve vertices
_ consider bringing back text/image using cache/names
_ exactly how pixel-filling works for strokes
_   http://processing.org/bugs/bugzilla/1025.html
_ deprecate loadPixels for 2.0? need to move away from it
_   mostly just change the examples to not use loadPixels() anymore
_ put a Hash<Something> class in there
_   will this work to sort by the values?
_   there was another concern as well..
_ save() should probably come out of image, ja?
