-
Notifications
You must be signed in to change notification settings - Fork 732
Merge v3.0/dev commits into v3.1/dev #890
Conversation
(cherry picked from commit 62f96ef)
(cherry picked from commit 0679184)
(cherry picked from commit 5056885)
(cherry picked from commit 0b8ab44)
(cherry picked from commit a4998ff)
(cherry picked from commit 2e6575a)
(cherry picked from commit 5f61b4b)
(cherry picked from commit 78e90af)
(cherry picked from commit efe9d44)
this is what docker uses (cherry picked from commit d8442ca)
(cherry picked from commit f1616dc)
(cherry picked from commit cbfc1a1)
(cherry picked from commit e3c6f85)
(cherry picked from commit 6b55d51)
(cherry picked from commit 4b8c060)
(cherry picked from commit 5856c18)
(cherry picked from commit 8cd8103)
(cherry picked from commit 7e894fa)
(cherry picked from commit 8bda0ca)
(cherry picked from commit e14d747)
(cherry picked from commit 249878a)
(cherry picked from commit b2ec5ca)
(cherry picked from commit 07bcc8f)
(cherry picked from commit 05d139e)
(cherry picked from commit 31de142)
(cherry picked from commit 6730214)
(cherry picked from commit 91ccb12)
(cherry picked from commit 9c78843)
(cherry picked from commit 855d292)
(cherry picked from commit ed9f133)
|
Wow. Thank you for this huge change set @lifeforms. You warned us of this growing gap and I am very happy you not only warned us, but you also stepped in to fix it for us. The change set was a bit less terrifying once I realized that most of it is regression testing. I have looked over the whole change set and then did a 2nd round looking at the conflicts. I am not totally sure I get it right, but I see diffs at the following two commits: Everything else looks good to me. |
|
Can these conflicts be solved and incorporated to this PR? |
|
I agree on the identical nature of the commits mentioned above. I am open to merge this at last, but the conflicts should be resolved first. Or should I try and resolve that? |
|
Fixing the conflicts; directly in the web editor. .travis.ymltrivial rules/REQUEST-933-APPLICATION-ATTACK-PHP.confThis is a change ( rules/REQUEST-943-APPLICATION-ATTACK-SESSION-FIXATION.confEscaping a dot. |
|
OK. I think we are ready to merge. Conflicts resolved. I do not think any additional drift in |
We are devving on v3.1/dev, but we are missing a large number of commits which were made on the v3.0/dev branch in the meantime. For instance, the regression tests and also some rule fixes were in v3.0/dev. I need to add tests so I made a PR that merges all stragglers that remained on v3.0/dev in one go.
After this merge, we should be very careful to not accept a change to v3.0/dev again (unless it's a backported fix). I hope the bot or some Github setting can help us keep that rule.
Reviewers
I hope this can be integrated quickly. But it's a huge merge. I'm not a Git wizard and I'm notoriously bad at resolving conflicts, although I did my best.
Please look extra closely at the commits I marked as
cp,conflictcause those involved manual work, oralready-doneas they seemed to be in v3.1 already so I skipped them.Extra point for @csanders-git: Could you verify that the regressions code was ported correctly? Maybe you can do a diff on the whole directory, cause the files should be the same in v3.1/dev. In case of discrepancies or things that need to be changed for v3.1/dev, you should be able to edit this PR.
What I did
A
git merge v3.0/devfailed catastrophically and gave an enormous list of conflicts in almost every source line, likely because of the indentation changes already done in v3.1/dev. (Should have done this before the indent changes, oh well.) It seemed enormously error-prone to go resolve all those conflicts manually or mass-ignore them. So I cherry picked commits from v3.0/dev to minimize and localize the conflict resolutions, at the cost of a bit messier Git history.I used
git log v3.0/dev ^v3.1/dev --no-merges --reverseto see which commits were supposedly only in v3.0/dev and not in v3.1/dev. Then I didgit showto investigate the old commit andgit cherry-pick -xto pick only the needed commits to v3.1/dev. It turns out that some of these commits were already there anyway, and some caused conflicts which I had to resolve.My log (
cp= cherry-picked,cp,conflict= cherry-picked and resolved conflicts,already-done= commit appeared already to be present in v3.1/dev,moot= commit is no longer necessary)