3.2.x - Replacing other 3.2.x merge - please review

[hazendaz]

A PR had been created to merge 3.2.x line back to master. The PR appeared to have lots of duplicate items. I took the time to perform two separate passes through a rebase to see if there were truely any changes. The original PR couldn't be merged to start with so I'm guessing over time 3.2.x was being cherry-picked from the master since it was really targetting 3.3.x rather than 3.2.x. At any rate, based on my two passes, there were only a couple of minor changes. I'm not sure these are even needed. I only skipped one real change with language specific modifications for zh doc that I personally cannot undertand therefore not sure if it was actually valid. Other than that only two items were valid. So I have created this PR in replacement of the other but want either @harawata or @kazuki43zoo to verify the two changes they made on 3.2.x to determine validity for going to the master. If this ends up being all junk, we should instead delete the 3.2.x branch.

For reference, the commit for zh doc that I skipped was this and it only applied to what I think is Chinese but I'm certainly no expert at all there. Most of that commit is already on the master.

0ae18b8

[hazendaz]

OK - of 3.2.x branch, commit listed as #376: Adjust the assertion of test case for 3.2.x line appaers to be invalid. If I recall correctly, internally mybatis is now using javassist not cglib and looking at that change it certainly appears that it is mixed up. So now just to address the ADD_MONTHS change from @harawata. To @harawata if your change here appears to not be necessary go ahead and close this off. I do think overall we probably could just delete 3.2.x branch and be ok.

[harawata]

I think this is about #399
Let me explain the background first.

We created 3.2.x branch so that we can release bug fix versions until 3.3.0 (which contains some backward incompatible changes) is released.
So, basically, we make all changes in the master and port only bug fixes to 3.2.x branch.
As we never port changes from 3.2.x to master, I think we can safely ignore #399 if that's what it tries to do.

Regarding the ADD_MONTHS change, I had to correct it in 3.2.x branch because its HSQLDB version was older than master.

[kazuki43zoo]

@harawata , thank you for explain the background of branch management policy.
In #376, i was need changing the assertion of test case because default byte code operation library(JAVASSIST or CGLIB) is different between master(3.3 line) and 3.2.x(3.2 line).
In other words, above 3.2.x changes cannot merge to master (unit test is failed).

[hazendaz]

So there we have it. I'm closing this PR out. Thanks everyone!

[hazendaz]

I updated the readme to markdown on 3.2.x branch and referenced this PR and #399

[harawata]

Thank you for your work @hazendaz !
I think #399 was a result of some mis-operation (we've received PRs like that from time to time).
I'm glad to have you in the team :-)

[hazendaz]

Thank you as well! Glad to be in the fold.