doing cheaper, probabilistic checks outside the scattergl conversion loop

Author: monfera

Diff for: src/plots/cartesian/axis_autotype.js

@@ -0,0 +1,73 @@
+/**
+* Copyright 2012-2016, Plotly, Inc.
+* All rights reserved.
+*
+* This source code is licensed under the MIT license found in the
+* LICENSE file in the root directory of this source tree.
+*/
+
+
+'use strict';
+
+var isNumeric = require('fast-isnumeric');
+
+var Lib = require('../../lib');
+var cleanDatum = require('./clean_datum');
+
+module.exports = function autoType(array) {
+    if(moreDates(array)) return 'date';
+    if(category(array)) return 'category';
+    if(linearOK(array)) return 'linear';
+    else return '-';
+};
+
+// is there at least one number in array? If not, we should leave
+// ax.type empty so it can be autoset later
+function linearOK(array) {
+    if(!array) return false;
+
+    for(var i = 0; i < array.length; i++) {
+        if(isNumeric(array[i])) return true;
+    }
+
+    return false;
+}
+
+// does the array a have mostly dates rather than numbers?
+// note: some values can be neither (such as blanks, text)
+// 2- or 4-digit integers can be both, so require twice as many
+// dates as non-dates, to exclude cases with mostly 2 & 4 digit
+// numbers and a few dates
+function moreDates(a) {
+    var dcnt = 0,
+        ncnt = 0,
+        // test at most 1000 points, evenly spaced
+        inc = Math.max(1, (a.length - 1) / 1000),
+        ai;
+
+    for(var i = 0; i < a.length; i += inc) {
+        ai = a[Math.round(i)];
+        if(Lib.isDateTime(ai)) dcnt += 1;
+        if(isNumeric(ai)) ncnt += 1;
+    }
+
+    return (dcnt > ncnt * 2);
+}
+
+// are the (x,y)-values in td.data mostly text?
+// require twice as many categories as numbers
+function category(a) {
+    // test at most 1000 points
+    var inc = Math.max(1, (a.length - 1) / 1000),
+        curvenums = 0,
+        curvecats = 0,
+        ai;
+
+    for(var i = 0; i < a.length; i += inc) {
+        ai = cleanDatum(a[Math.round(i)]);
+        if(isNumeric(ai)) curvenums++;
+        else if(typeof ai === 'string' && ai !== '' && ai !== 'None') curvecats++;
+    }
+
+    return curvecats > curvenums * 2;
+}

Diff for: src/plots/cartesian/axis_defaults.js

@@ -23,9 +23,8 @@ var handleTickLabelDefaults = require('./tick_label_defaults');
 var handleCategoryOrderDefaults = require('./category_order_defaults');
 var setConvert = require('./set_convert');
 var orderedCategories = require('./ordered_categories');
-var cleanDatum = require('./clean_datum');
 var axisIds = require('./axis_ids');
-
+var autoType = require('./axis_autotype');
 
 /**
  * options: object containing:
@@ -207,13 +206,6 @@ function isBoxWithoutPositionCoords(trace, axLetter) {
     );
 }
 
-function autoType(array) {
-    if(moreDates(array)) return 'date';
-    if(category(array)) return 'category';
-    if(linearOK(array)) return 'linear';
-    else return '-';
-}
-
 function getFirstNonEmptyTrace(data, id, axLetter) {
     for(var i = 0; i < data.length; i++) {
         var trace = data[i];
@@ -228,54 +220,3 @@ function getFirstNonEmptyTrace(data, id, axLetter) {
         }
     }
 }
-
-// is there at least one number in array? If not, we should leave
-// ax.type empty so it can be autoset later
-function linearOK(array) {
-    if(!array) return false;
-
-    for(var i = 0; i < array.length; i++) {
-        if(isNumeric(array[i])) return true;
-    }
-
-    return false;
-}
-
-// does the array a have mostly dates rather than numbers?
-// note: some values can be neither (such as blanks, text)
-// 2- or 4-digit integers can be both, so require twice as many
-// dates as non-dates, to exclude cases with mostly 2 & 4 digit
-// numbers and a few dates
-function moreDates(a) {
-    var dcnt = 0,
-        ncnt = 0,
-        // test at most 1000 points, evenly spaced
-        inc = Math.max(1, (a.length - 1) / 1000),
-        ai;
-
-    for(var i = 0; i < a.length; i += inc) {
-        ai = a[Math.round(i)];
-        if(Lib.isDateTime(ai)) dcnt += 1;
-        if(isNumeric(ai)) ncnt += 1;
-    }
-
-    return (dcnt > ncnt * 2);
-}
-
-// are the (x,y)-values in td.data mostly text?
-// require twice as many categories as numbers
-function category(a) {
-    // test at most 1000 points
-    var inc = Math.max(1, (a.length - 1) / 1000),
-        curvenums = 0,
-        curvecats = 0,
-        ai;
-
-    for(var i = 0; i < a.length; i += inc) {
-        ai = cleanDatum(a[Math.round(i)]);
-        if(isNumeric(ai)) curvenums++;
-        else if(typeof ai === 'string' && ai !== '' && ai !== 'None') curvecats++;
-    }
-
-    return curvecats > curvenums * 2;
-}

Diff for: src/traces/scattergl/convert.js

@@ -17,6 +17,7 @@ var isNumeric = require('fast-isnumeric');
 
 var Lib = require('../../lib');
 var Axes = require('../../plots/cartesian/axes');
+var autoType = require('../../plots/cartesian/axis_autotype');
 var ErrorBars = require('../../components/errorbars');
 var str2RGBArray = require('../../lib/str2rgbarray');
 var truncate = require('../../lib/float32_truncate');
@@ -261,6 +262,29 @@ proto.update = function(options) {
     this.color = getTraceColor(options, {});
 };
 
+// We'd ideally know that all values are of fast types; sampling gives no certainty but faster
+//     (for the future, typed arrays can guarantee it, and Date values can be done with
+//      representing the epoch milliseconds in a typed array;
+//      also, perhaps the Python / R interfaces take care of String->Date conversions
+//      such that there's no need to check for string dates in plotly.js)
+// Patterned from axis_defaults.js:moreDates
+// Code DRYing is not done to preserve the most direct compilation possible for speed;
+// also, there are quite a few differences
+function allFastTypesLikely(a) {
+    var len = a.length,
+        inc = Math.max(0, (len - 1) / Math.min(Math.max(len, 1), 1000)),
+        ai;
+
+    for(var i = 0; i < len; i += inc) {
+        ai = a[Math.floor(i)];
+        if(!isNumeric(ai) && !(ai instanceof Date)) {
+            return false;
+        }
+    }
+
+    return true;
+}
+
 proto.updateFast = function(options) {
     var x = this.xData = this.pickXData = options.x;
     var y = this.yData = this.pickYData = options.y;
@@ -272,18 +296,18 @@ proto.updateFast = function(options) {
         pId = 0,
         ptr = 0;
 
-    var xx, yy, fastType;
+    var xx, yy;
+
+    var fastType = allFastTypesLikely(x);
+    var isDateTime = !fastType && autoType(x) === 'date';
 
     // TODO add 'very fast' mode that bypasses this loop
     // TODO bypass this on modebar +/- zoom
     for(var i = 0; i < len; ++i) {
         xx = x[i];
         yy = y[i];
 
-        // check for isNaN is faster but doesn't skip over nulls
-        fastType = isNumeric(xx) || xx instanceof Date;
-
-        if(isNumeric(yy) && (fastType || Lib.isDateTime(xx))) {
+        if(isNumeric(yy) && (fastType || isDateTime)) {
 
             if(!fastType) {
                 xx = Lib.dateTime2ms(xx);