Use package dependencies for ES6 Array shims #962
Conversation
The shims that are vendored in to the library break in IE; also, no reason to reinvent the wheel. Instead, use the `array.prototype.find` and `array-includes` modules to ensure compatibility.
outoftime
force-pushed
the
array-shims
branch
from
bc03645
to
dc61d57
Compare
|
LGTM. We should figure out why our unit tests didn't catch this. |
The shims that are vendored in to the library break in IE; also, no reason to reinvent the wheel. Instead, use the `array.prototype.find` and `array-includes` modules to ensure compatibility.
…ebid-official-0.19.0 to release/1.14.0 * commit 'b13f7ba7ee8b3c168dc0af7c8bfc94747d017e70': (34 commits) Add changelog entry. Prebid 0.19.0 Release check truthiness of adUnitCode (prebid#990) fixed exception when refreshing individual Ad Units and bidder responds slowly (prebid#989) Stub pixel drop to prevent network request in test (prebid#988) Updating Komoona adapter to support future Prebid requirements (prebid#974) Use stable version of Chrome (prebid#984) Revert to running browser tests in Travis (prebid#983) Fix issue with appnexusAst sending `user` object in the wrong place. (prebid#980) Integrate Browserstack tests into Travis CI build (prebid#839) Add StickyAdsTV Bidder adapter (prebid#916) Stronger xdomain checks (prebid#971) added matomy as an alias for appnexus (prebid#850) Added 152Media Appnexus Alias (prebid#952) OpenX Adapter: Handles fallback ads correctly as a no fill (prebid#39) (prebid#963) Use package dependencies for ES6 Array shims (prebid#962) Make x-domain safe frame example work out of the box (prebid#955) added usersync for adkernel adapter (prebid#951) Rubicon adapter: add a floor variable (prebid#964) Added Lifestreet adapter. (prebid#965) ...
…14.0 to master * commit 'c008f3f531ae3409f4a16bf03470d84e82aead0e': (35 commits) Add adapters in aolPartnersIds.json. Add changelog entry. Prebid 0.19.0 Release check truthiness of adUnitCode (prebid#990) fixed exception when refreshing individual Ad Units and bidder responds slowly (prebid#989) Stub pixel drop to prevent network request in test (prebid#988) Updating Komoona adapter to support future Prebid requirements (prebid#974) Use stable version of Chrome (prebid#984) Revert to running browser tests in Travis (prebid#983) Fix issue with appnexusAst sending `user` object in the wrong place. (prebid#980) Integrate Browserstack tests into Travis CI build (prebid#839) Add StickyAdsTV Bidder adapter (prebid#916) Stronger xdomain checks (prebid#971) added matomy as an alias for appnexus (prebid#850) Added 152Media Appnexus Alias (prebid#952) OpenX Adapter: Handles fallback ads correctly as a no fill (prebid#39) (prebid#963) Use package dependencies for ES6 Array shims (prebid#962) Make x-domain safe frame example work out of the box (prebid#955) added usersync for adkernel adapter (prebid#951) Rubicon adapter: add a floor variable (prebid#964) ...
|
@outoftime Do you recall what version of IE? I'm testing our site in IE 9 and it works fine in Prebid 0.18 without this PR. Could this have been something else in your app? |
|
Good point, the screenshot references another script file and we test extensively in IE. |
|
The problem was observed in IE 11. I can tell you that the change in this PR, in isolation, fixed the error. You’re correct that the error was being thrown from our main application bundle, so it’s possible that the original polyfill differed subtly from real browser behavior in terms of the property descriptor for the polyfilled methods, which caused another library in our bundle to break. Anyway, at this point we bundle Prebid into a JS bundle that we control, so if you roll this back we can still work around the problem by importing an appropriate polyfill before importing Prebid. |
|
We are gonna to roll back this change, however, I'm not sure it's causing the issue you think it is. I just ran this code in IE 11: https://jsfiddle.net/tn3d87zy/ and don't see any exception. Also - note that the same code is used in the polyfill includes as it was with the manual includes we had (it does a check on |
|
@mkendall07 yeah it’s hard to say at this point, although maybe I can get more info on the error once this is rolled back. I think it’s probably more subtle than a transparent error in redefining |
The shims that are vendored in to the library break in IE; also, no reason to reinvent the wheel. Instead, use the
array.prototype.findandarray-includesmodules to ensure compatibility.Original error message in IE: