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 Nov 6, 2023. It is now read-only.
Repository navigation
This repository was archived by the owner on Nov 6, 2023. It is now read-only.
@J0WI thank you for opening this issue. I would like to add some details here.
In #10869 I checked if a target is valid by matching it against the Public Suffix List. While this test does not guarantee each target has a working DNS record, it has the advantage that no network request is required to perform the checking.
It was found that AM-Best-Company.xml#L11 with a invalid target passed the comprehensive fetch test and got re-activated in #10538. Despite @Bisaloo's assertion, I would like to hear from @Hainish for an explaination.
@Hainish@J0WI I have implemented the logic which check each target is having a valid gTLD/ ccTLD in the rule-test in #11187. Besides, I noted that the following type of rewrite will pass the fetch-test without any Travis complaint
@Hainish For any target if HTTPS connection works, the fetch test will ignore any result for the HTTP connection, see check_rules.py#L131. In particular, it will ignore the could not resolve host error. As a result, any rewrite in the form of #10877 (comment) will always pass the fetch test.
It seems to be a bug to me. A quick fix is to throw an error if HTTP cannot be resolved. Can you confirm this? Besides, what should be done for ip target as in Xxxbunker.com.xml#L44?
All ruleset in the code base will pass the proposed rule-test once #11263 is merged, it will be also safe to merge #11187 (hopefully before next release). See also #11264.
Updated #11187 to include the changes in #11264. thanks!
Activity
cschanaj commentedon Jul 6, 2017
@J0WI thank you for opening this issue. I would like to add some details here.
In #10869 I checked if a
targetis valid by matching it against thePublic Suffix List. While this test does not guarantee each target has a working DNS record, it has the advantage that no network request is required to perform the checking.It was found that
AM-Best-Company.xml#L11with a invalidtargetpassed the comprehensive fetch test and got re-activated in #10538. Despite @Bisaloo's assertion, I would like to hear from @Hainish for an explaination.cschanaj commentedon Jul 16, 2017
@Hainish @J0WI I have implemented the logic which check each
targetis having a validgTLD/ ccTLDin therule-testin #11187. Besides, I noted that the following type of rewrite will pass thefetch-testwithout any Travis complaintIs this an intended behavior? Should we check each
targetfor an existing DNS record? See cschanaj/https-everywhere/pull/6Besides, should we allow IP
targetas inXxxbunker.com.xml#L44?Please be noted that the code does not compile currently.cschanaj commentedon Jul 19, 2017
@Hainish can you have a look on #10877 (comment) and #11187 ?
Hainish commentedon Jul 20, 2017
Not quite sure how
AM-Best-Company.xml#L11got re-enabled. Checking targets against the Public Suffix List seems reasonable.cschanaj commentedon Jul 20, 2017
@Hainish For any
targetif HTTPS connection works, the fetch test will ignore any result for the HTTP connection, seecheck_rules.py#L131. In particular, it will ignore thecould not resolve hosterror. As a result, any rewrite in the form of #10877 (comment) will always pass the fetch test.It seems to be a bug to me. A quick fix is to throw an error if HTTP cannot be resolved. Can you confirm this? Besides, what should be done for ip
targetas inXxxbunker.com.xml#L44?cschanaj commentedon Jul 21, 2017
All ruleset in the code base will pass the proposed
rule-testonce #11263 is merged, it will be also safe to merge #11187 (hopefully before next release).See also #11264.Updated #11187 to include the changes in #11264. thanks!
Update https_everywhere_checker/check_rules.py, Fix #10877 (#11187)