You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Apr 17, 2026. It is now read-only.
Repository navigation
This repository was archived by the owner on Apr 17, 2026. It is now read-only.
Hi, I have one problem. When I'm trying to download the latest chromedriver:
npx webdriver-manager update --gecko=false --standalone=false --versions.chrome=85.0.4183.83
the version which is downloaded is 85.0.4183.38.
Do you have any idea, why it not download the correct version?
I/downloader - curl -oC:\TEST\Sources\Src\TestApps\node_modules\webdriver-manager\selenium/chromedriver_85.0.4183.83.zip https://chromedriver.storage.googleapis.com/85.0.4183.38/chromedriver_win32.zip
We are facing this issue too, as 85.0.4183.87 has some fixes to deal with closing alerts that we need. Unfortunately, webdriver-manager update uses the first version it can find for 85.0.4183 (which is .38), rather than the last version (.87). This is happening somewhere in the getSpecificChromeDriverVersion() function of chrome_xml.ts.
While debugging with @Blackbaud-TerryHelems, it seems this bug has been present for quite some based on #424 and #408. I believe it is only applicable on win32 systems using an x64 architecture where multiple versions exist for the same major version. The reason being line 103.
As we iterate through the (already filtered to the OS) list of possible versions, the code is able to catch the last matching version (with the last .digits removed) because those fall in to the successful else if branch as mac64 and linux64 exist. In windows; however, that second nested if fails since no versions contain win64 (so we're stuck with the first match).
I used the red -> green approach with the unit test I added in #473 (but forgot to do separate commits :-)) to show the current bug and potential fix.
Has anyone found a reliable way to work around this issue? I've tried downloading the correct version of ChromeDriver directly, and placing it in the correct node_modules sub-folder inside the protractor dependency, but it just gets overwritten later as Protractor itself updates it pretty late.
Some other options I can think of:
Patch the affected chrome_xml.js file on the fly.
Start a Selenium server and configure Protractor to use that.
Activity
Tyf0x commentedon Aug 31, 2020
Does anybody else have this issue? This broke our e2e tests on appveyor.
We're trying to use
webdriver-manager update --versions.chrome $env:CHROME_VERSION --verbose --gecko falsewhere CHROME_VERSION = 85.0.4183.87webdriver-manager ends up downloading the version 85.0.4183.38 instead BUT saving it in a archive named 85.0.4183.87.
This then ends up breaking the tests because chromedriver cannot find chrome:

Which I assume was fixed in build .83:
Shanson90 commentedon Sep 1, 2020
We have this issue as well.
lotz commentedon Sep 1, 2020
We are facing this issue too, as
85.0.4183.87has some fixes to deal with closing alerts that we need. Unfortunately,webdriver-manager updateuses the first version it can find for85.0.4183(which is.38), rather than the last version (.87). This is happening somewhere in thegetSpecificChromeDriverVersion()function ofchrome_xml.ts.Blackbaud-BobbyEarl commentedon Sep 2, 2020
We're seeing the same issue.
While debugging with @Blackbaud-TerryHelems, it seems this bug has been present for quite some based on #424 and #408. I believe it is only applicable on
win32systems using anx64architecture where multiple versions exist for the same major version. The reason being line 103.As we iterate through the (already filtered to the OS) list of possible versions, the code is able to catch the last matching version (with the last
.digitsremoved) because those fall in to the successfulelse ifbranch asmac64andlinux64exist. In windows; however, that second nested if fails since no versions containwin64(so we're stuck with the first match).I used the red -> green approach with the unit test I added in #473 (but forgot to do separate commits :-)) to show the current bug and potential fix.
lotz commentedon Sep 4, 2020
Has anyone found a reliable way to work around this issue? I've tried downloading the correct version of ChromeDriver directly, and placing it in the correct
node_modulessub-folder inside theprotractordependency, but it just gets overwritten later as Protractor itself updates it pretty late.Some other options I can think of:
chrome_xml.jsfile on the fly.lotz commentedon Sep 11, 2020
I've worked around this by patching the
chrome_xml.jsfile on the fly (using a super hacky approach): https://gist.github.com/lotz/d33f550d48dd125344d62321a8905d8bTo use, just add to your NPM
postinstallscript like so: