Merge avr-100-pin branch into master #49
Conversation
This branch is a work in progress. The intention is to add support for the ATmega640/1280/2560 to this Arduino core as well.
Delete duplicate file. Located in the src folder instead
Delete duplicate file. Located in the src folder instead
It's easier to maintain one copy of optiboot, rather than four. That's why I've merged optiboot_flash from MightyCore, MiniCore, MajorCore and ButterflyCore into this one. Hopefully this will be helpfull when the official optiboot testing will begin
Needed to fix the new bootloader path + pinout naming
More self explaining names where needed
git-subtree-dir: avr/bootloaders/optiboot_flash git-subtree-split: ab5e7651397c20b1caa32a8139327af93c5b4a0d
…oaders/optiboot_flash'
Correctly handle failure at all build lifecycle steps. This should allow the report push to be retried if it fails and if the retries fail it will still complete the rest of the build and not cause a failure. per1234/arduino-ci-script@6ec96b8
Now that error handling in the script has been fixed, Travis CI will be able to track job success based on the return value of build_sketch so it's no longer necessary to have a function for this.
Update arduino-ci-script subtree to 6ec96b8
Fix bug that causes build_sketch to return false negatives when: - build_sketch arguments specify compilation of a single sketch with multiple IDE versions. - One of the complations fails - The last compilation before build_sketch returns is successful per1234/arduino-ci-script@dc0c81d
Retry verification after arduino returns any undocumented error code. This will provide multiple opportunities for a successful sketch compilation in the event that a temporary glitch unrelated to the sketch or script causes a failture. per1234/arduino-ci-script@7fd36de
Update arduino-ci-script for correct failure handling
Without the leading zeros the folder listing in the report repository is not correctly ordered.
Add leading zeros to the report folder name
|
I need some help here. I tried to resolve this issue from the command line, but git added It seems like the conflicting files is |
|
There are some conflicts as the history of the branches are combined because they have diverged a little. I just had a try at resolving them but I was tripped up somehow by the subtree commits so I need to redo it. All the commit history from the avr-100-pin branch should be added to master during the merge. I feel like these should be rebased onto it so that none of the history of the master branch is changed. The one I saw that became redundant was c6ae0fd because, other than breaking line 70 into two lines, that change had already been made on the master branch at 30723e0 so I squashed that change into e546d43. I'm a little too tired now to make sure I don't mess this up but I'll have another try in the morning if you haven't fixed it by then. Resolving merge conflicts always make my head hurt. I always keep my finger crossed when I'm doing a rebase in hopes that I won't have any to deal with. |
|
Just wanted to give an update on my progress (or lack thereof). It turns out that for some crazy reason Git puts subtrees into the repository root when you rebase. I couldn't figure out why that kept happening to me. The only workaround I've found is that all the subtree merges need to be omitted from the rebase and then added back in one at a time: |
|
Well, this is what I've managed: The main thing I don't like is that the author field for every one of your commits says "MCUdude committed with per1234" now. I don't know how to get rid of that. I didn't end up exactly reproducing your subtree commits for optiboot_flash and MCUdude_corefiles. We must have a different system for doing that because mine happen in only a single commit but yours happen in multiple commits. I follow these instructions: |
I have no problems with you getting you name all over this merge. After all, you're been incredibly helpful. Does this mean you would have to create a new pull request, where you've merged avr-100-pin into the master of your fork, and then I merge your fork into this repos master, right? |
That's correct. I would just need to submit a pull request to merge my branch: I just need to wait for the Travis CI build of my branch and then I can submit the pull request. Unfortunately I just noticed that the build was blocked from starting by an ongoing build of a previous attempt that I had replaced but forgot to cancel so it's just getting started now: |
MCUdude
force-pushed
the
avr-100-pin
branch
3 times, most recently
from
bafb54e to
0485272
Compare
|
I got som help from a colleague, which is much more experienced with git than I am. We were able to merge the two branches, and created a new one, |
|
My work is more clean but go with whatever you're happy with. |
|
I'd allways prefer the cleanest. How is yours cleaner then |
|
I redid the subtree commits after the rebase so that they correctly place the subtrees in the correct subfolder of the repository, as the original commits did. Your subtree commits put the subtrees in the root folder, which required you to later add the commits moving them to the correct folder: The subtrees initially being added to the root in your The final result is essentially the same in either branch so either way you go it will be fine. I get kind of uptight about commit histories because I end up digging through them quite often. Mostly this is because I don't really know what I'm doing and I have to try to "reverse engineer" the purpose of some code that would probably be obvious to someone with more knowledge. When find the commit where that line was added or modified and it's a some non-atomic changes with a commit message that only says "Update foo.c" it makes me want to tear my hair out! |
|
Yep, my colleagues fix messes up the subtrees. If you create a PR, i'll merge it into the master |
I've been working on this branch for a while now. Support for ATmega640/1280/2560 is added, and Optiboot has gotten a major cleanup and is now maintained from its own repository. The Arduino core files now supports all my Arduino cores, and is also maintained from its own repository. Travis CI is now implemented, which makes testing and bug smacking much easier.