Chrome and Gecko distribute their drivers through some non-standard channels. All our autoinstaller code still works, we just aren't getting the archives from PyPi.
<rdar://problem/70551255>
Created attachment 412050 [details] Patch
Created attachment 412108 [details] Patch
Both of these patches work, I'm just not sure which approach I like better. One is more clear that web drivers are not actually pip packages, at the cost of making accessing the web drivers less portable, while the other pretends that web drivers are pip packages in exchange for making them portable.
Comment on attachment 412050 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=412050&action=review > Tools/Scripts/webkitpy/autoinstalled/chromedriver.py:37 > + response = requests.get(self.URL + 'LATEST_RELEASE') Should we handle the case when requests.get throws an exception? > Tools/Scripts/webkitpy/autoinstalled/chromedriver.py:54 > + link='{url}{version}/chromedriver_{os}.zip'.format( How about making it one-liner, like: link = '{}{}/chromedriver_{}.zip'.format(self.URL, version, os) > Tools/Scripts/webkitpy/autoinstalled/chromedriver.py:56 > + version=str(version), str might not be required here with format()
rs=me
Created attachment 412133 [details] Patch for landing
Committed r268896: <https://trac.webkit.org/changeset/268896> All reviewed patches have been landed. Closing bug and clearing flags on attachment 412133 [details].