jrburke (James Burke)

Breaking out the get() functionality as a separate function to allow easier overriding for another host environment.

Get enough of the modules to work to run the uri tests. Not all of them pass, at least one, the fragment one does not seem possible to pass since the uri module does not have a reference to fragment, maybe the test is out of date.

Basic conversion script for CommonJS modules. May be hard to use in practice if the CommonJS modules were written with lots of custom path entries. Tried it on narwhal/lib, but those files seem to assume some very specific module paths for some prefixes.

Adding support for exports and module for CommonJS converted modules. Now need to do a conversion script and add some tests.

Now that run can be configured to load some modules as soon as it is defined, found a weakness with the convert script where the dojo/_base modules were not being defined at the right time.

Support config values for deps, callback and ready, to specify run() calls before run.js is loaded. Updated README.

Automating the process of updating the jquery sample to make a distributable for it.

Changing i18n bundles to just list the bundles, not full names, since it is redundant, based on suggestion from Adam Peller.

Update run integration to use run.def to define modules.

Fix issue where loaded resources without names would not get called back correctly. Introduced test that demonstrates the issue and the fix.

Fixing some bugs in the re-tooling of toggleClass, adding in some performance optimizations.

Fix a bug with run.modify() where it was not tracking the name correctly for a modifier in the case where hte modify call defines a module/function callback, and make sure run.ready() waits for isDone, indicating all modules have been evaluated.

Rework the evaluations of module callbacks, using a recursive execute style as suggested by Rawld Gill. This style was not used before in the dojo xd loader because it had to support Safari 2 which had a tiny call stack and the recursive calls would kill it easily. Should not be a problem now for the browsers of today. Had to rework the build process to account for new exec method. Added the jquery-run-sample to source to make it easier to track, but needs a tweak from the source config to make it usable standalone, as mentioned in the README.txt in that folder.

Fixes issue #2, a typo and stronger config.optimize string check. Thanks rcgill

Renamed run.convertNameToPath to be nameToUrl and placed contextName as the third arg, to match other context-aware methods. Also exposed it on the context-specific run method for use in modules.

Tweaks suggested by Bryan Forbes/David Mark: avoiding this refs, using .src to get absolute path for urls found through script element checking, use most recent script tag parentNode if head element is not available, remove global and doc refs and not use this, but favor a var declared outside the master closure, stricter loaded/complete regexp. Added a test page that has no html, head or body tags in it.

Fix a break issue when looking for run.js script tag, where if run is added not via inline script src tag there could be a problem, also pass null explicitly to xhr.send to make FF 3.0 happy with fetching text! resources.

Set the url at the end for good measure, for IE's sake, recommended by Bryan Forbes.

Refs Issue #1, clarify the need for nls in the i18n names, and make sure the default case in the i18n.html test file specifies the i18n! prefix.

Making license blob more explicit for each piece, so when closure compiler aggregates them at the top it is clear what the license refers to, also fixed some things with passing config via a run = {} object, and removed some old cruft from the modifer.html test page.

Set up build system to be able to handle dojo files with pragmas in it, and allow run.js to be built into dojo.js and have the baseUrl configured correctly and remove the use strict for dojo layers that go through closure compiler, since that compiler uses different rules with that directive.

Fix issue where files were being added multiple times to a build layer because the added files were not persisted across callModules, which is called more often in the sync loader case.

Fix dojo conversion script to convert module names to be slash separated, and removed the old back-compat shim in run.js for it.

Try to clarify a bit what blade hopes to be. Related to issue #1

Bringing Sizzle test suite changes back in.

link updates

One possible approach to integrating RunJS with jQuery.

Only complain about expiry if something has not loaded.

Updating jquery usage doc.

Add an appDir option to the build, to have a complete app copied over, but then treat the baseUrl as relative subdirectory to that app path.

Fix circular tests so that the modules work no matter what order they are asked for, some of the modules were not using run.get() yet.

Make sure files that do not have a run() call with a module name get one when building a layer, so that the layer does not try to load the file later after run.resume().