Add macOS CI #120
Add macOS CI #120
Conversation
|
Yeah, I'll definitely need @tsjensen to help with this. I know I'm doing something dumb... :) |
|
Well, I've gotten closer. First, I had to install the GNU versions of sed, grep, and xargs since the BSD ones that macOS has by default are odd and your tests use GNU options (and I don't mind this because I hate the BSD versions of those utilities). But even with that, as far as I can tell the white-box tests can't run at the moment since macOS ld does not support --wrap and I'm not sure how to get around that. Frankly, I'm not sure what it's doing. 😄 Now the black-box tests get pretty much through everything...but dies at the end with an error. I can see one test fails for Unicode/language-y ( Internet said setting I also see: and then finally: |
|
That's great progress! Here's my 2 cents on some of the topics:
|
|
Don't know if this is helpful, or if you've already worked this out, but on my M1 Mac Mini, I tried to figure out if there was a way to simulate the Finally, I decided to try the "nuclear option", and was able to get the unit tests to pass by defining a function pointer (I also added a section in flags_osx:
$(eval CFLAGS := -I. -I$(SRC_DIR) -O -Wall -W -Wno-stringop-overflow $(CFLAGS_ADDTL))
$(eval LDFLAGS := $(LDFLAGS) --coverage $(LDFLAGS_ADDTL))
$(eval UTEST_EXECUTABLE_NAME := unittest)
$(eval UTEST_OBJ := $(UTEST_NORM:.c=.o))If this is helpful, I'd be glad to submit a pull request with my changes (or maybe an update this pull request? ... I'm not sure of the proper way to do this kind of thing). Here's the output of |
|
Incidentally, I haven't yet merged in @mathomp4's changes from this mathomp4:add-macos-ci branch -- I created my branch off of the color-unicode branch, since the whole reason I wanted to build from source was to get ANSI box drawing characters to work... which they do (although the terminal font has a big effect on how nice they end up looking): $ echo "boxes is great!" | out/boxes -d ansi
┌─────────────────┐
│ boxes is great! │
└─────────────────┘
$ echo "boxes is great!" | out/boxes -d ansidouble
╔═════════════════╗
║ boxes is great! ║
╚═════════════════╝
$ echo "boxes is great!" | out/boxes -d ansirounded
╭─────────────────╮
│ boxes is great! │
╰─────────────────╯
$ echo "boxes is great!" | out/boxes -d ansiheavy
┏━━━━━━━━━━━━━━━━━┓
┃ boxes is great! ┃
┗━━━━━━━━━━━━━━━━━┛
$ echo "boxes is great!" | out/boxes -d ansidashed
┌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌┐
┊ boxes is great! ┊
└╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌┘
$ |
|
@chorpler That looks very promising. Thank you for sharing! We would much appreciate you contributing this. It seems to me that your changes and @mathomp4's should work together nicely. Maybe you could just update this PR? I would then rebase the color-unicode branch on top of it, so we can already build it on Mac. |
|
It might be that updating this PR is not so straightforward because its content is from another fork. So, please feel free to open another PR if that's easier for you. 🙂 |
|
@chorpler I'm willing to do whatever you'd like to get this in. You can either fork my fork and make a PR to my branch. Or, since my changes are so small, if you make a PR with your changes, I can merge your branch into this branch and we can see if the macos CI works with our combined power! |
|
Had some "fun" and added a bit of ugly bash to work around test 111 on macOS. Also fixed up the coveralls since we now have two checks instead of one. I also added an "ignore error" for the issue that will probably be fixed by the work of @chorpler . But I wanted to make sure the full workflow would work if we had everything working. 😄 |
|
Last night I tried to figure out the best way to add my changes. My first attempt at cherry-picking commits to merge into @mathomp4's branch ended up pulling in some changes from the color-unicode branch (from the same files I'd made changes in) and rendered the whole thing completely unbuildable. :D I ended up creating a branch from the mathomp4:add-macos-ci branch and just redoing my changes on top of it. Which was good, because I realized that if I called the function pointer Then I checked that things would still build and run on Linux and Windows.
Anyway, I'll go ahead and create the pull request to merge my new branch into the existing mathomp4:add-macos-ci branch, and we'll go from there. |
|
I created a pull request here: mathomp4#1 Hopefully that will allow this pull request to be updated also. If it doesn't work, we can create a new pull request. |
|
I've pulled in the branch from @chorpler into my branch. Good news: macOS passes both white-box and black-box tests! 🎉 Bad news: Linux is now failing white-box with segfaults and other stuff 😞 : https://github.com/ascii-boxes/boxes/actions/runs/6997063497/job/19033607396 I even tried adding: env:
LEX: flex
YACC: bisonto the action in case I hit the Fedora issue @chorpler saw above, but it didn't change anything. I'm a Fortran programmer by trade so I'm not great at debugging C. Hopefully you all can see what's happening! |
|
OK, yeah, I see what I did. While redoing my changes in the I've submitted a new pull request for your mathomp4:add-macos-ci branch, @mathomp4. It looks like the |
|
It occurs to me that as an alternative to changing line in |
|
Undrafted! 🎉 |
|
I would appreciate if you removed the |
|
Sounds good, I'll remove it. Thanks for the tip on Windows bison using |
tsjensen
force-pushed
the
add-macos-ci
branch
from
21d7a90 to
81103fc
Compare
|
One more thing, a bit about cosmetics ... I am a big fan of a linear commit history in my projects (no merge commits). So, I would like to do some cleanup of the commit history on this branch before I merge it, also rebasing it onto the latest master and so on. (Content will not be changed, of course.) I would offer to do that myself, but only if it's OK with you. 😊 If you like, I can also merge mathomp4#3 while I'm at it. |
Huh. I didn't see a new PR. Good job GitHub notifications. Let me do that real quick. And then after that, it's up to you. But first, merge time... |
|
Okay. The PR from @chorpler is merged in. |
…`__wrap_bx_fprintf()`
Remove flags_darwin, which are the same as flags_unix now
tsjensen
force-pushed
the
add-macos-ci
branch
from
858044a to
f1bda75
Compare
|
Ok, now we have nice, linear commits. 😸 I also made another small change regarding the defaults for |
This is attempting to add macOS in the CI.