Improve performance and fix issues #728
Conversation
1. Add htmlButtonText option 2. Add htmlOptionText option 3. Use more local variables instead of frequently referencing widget properties/options. 4. Use document.createElement() to more quickly create new DOM elements. 5. Eliminate use of href="#" in header links. Add cursor: pointer to jquery.multiselect.css for these links. 6. Move insertions to end of _create to limit browser reflowing lags. 7. Use more ternary logic to eliminate if/then's where it makes sense. 8. Create _updateCache() internal function... called from multiple places. 9. Ditch use of $.isFunction. A simple typeof can be reliably used instead. 10. Add check in update() to reposition menu if button height has changed from saved value and the menu is open. 11. Save the current button height in position(). Tests Updated 12. Added test for namespace separation to core.js 13. Added selectedList:0 to tests in core.js to ensure that button text is correct for tests. 13. Fixed minWidth percentage test in options.js that was failing for non-Chrome browsers. Files updated: jquery.multiselect.js jquery.multiselect.filter.js jquery.multiselect.css core.js options.js
|
Thanks for fixing that test, I'm not sure when it started failing but I noticed it on your first PR. I poked at it a couple of times but hadn't had time to sit down and investigate it thoroughly. |
|
When I looked into the test, it appears that the difference is always 1 or less. You can get a difference of 1 if the parent div has an odd number of pixels width... This is why I thought that the floating point round off rationale is plausible. I just fixed the test by making a difference of 1 or less passable. |
|
I used to always see a difference of less than one which is why I added that rounding operation. On my mac I was seeing a difference of about 12 - 20. The issue was coming up because the element's parent width was changing substantially between lines of code in the test. I'll pull down your changes and see if it is still fixed. If it's not, your changes are still an improvement for other browsers and the onus isn't on you to fix specs that have decide to die on some browsers so I'll still merge it. |
|
Getting accurate reliable widths is one area where jQuery has historically had a number of bugs, so it may well be that this is a jQuery related issue. I would recommend keeping that in mind. In some cases, it may be necessary to ditch jQuery because of particular features that have had issues in the past. |
|
The extra commits are just minor stuff I missed. Oops! Trying to keep the commits as close as possible to what I actually have in my editor. ;-) |
Tests Updated
12. Added test for namespace separation to core.js
13. Added selectedList:0 to tests in core.js to ensure that button text is correct for tests.
13. Fixed minWidth percentage test in options.js that was failing for non-Chrome browsers--this is likely due to floating point round off differences between browsers.
Files updated:
jquery.multiselect.js
jquery.multiselect.filter.js
jquery.multiselect.css
core.js
options.js