imake: add imake/1.0.8 recipe

[madebr]

Specify library name and version: imake/1.0.8

For #6387

dependency tree: sdl -> nas -> imake -> xorg-proto -> xorg-macros

---

• I've read the guidelines for contributing.
• I've followed the PEP8 style guides for Python code in the recipes.
• I've used the latest Conan client version.
• I've tried at least one configuration locally with the
  conan-center hook activated.

[madebr]

@jgsogo
Can you (manually) start a job to create armv8 macos packages for xorg-proto?
xorg-proto was merged today. So it feels strange to open a pr on the same day to trigger creation of new packages.

[jgsogo]

xorg-proto fails for Macos+armv8: https://c3i.jfrog.io/c3i/misc/summary.html?json=https://c3i.jfrog.io/c3i/misc/logs/TapaholesList/232/summary.json

[jgsogo]

we are on it: #6686

[madebr]

This is probably because I added test_type = "build_requires" to the test package's recipe.

[madebr]

@jgsogo
The error #6677 (comment) shows that there is a possible bootstrap problem on c3i when cross building.

[madebr]

@jgsogo
Weird.
The log contains:

    imake/1.0.8:c58972ce7fa79168bcca07b3885d2a61402d8743 - Download

Where it should be

    imake/1.0.8:c58972ce7fa79168bcca07b3885d2a61402d8743 - Build

See #6677 (comment)

[jgsogo]

If you have a look at the output table, you can see it says UPSTREAM + TEST_FAILED for this configuration: the package is already available in the repository created for this pull-request (it was already built, latest commit is just modifying test_package), and we are just running conan test in the CI, so the package needs to be downloaded and the error comes from the conan test output, not from conan install --build.

The table is useful to know where the error comes from:

...although I agree the error message in GitHub should provide that information as well. Maybe we can just add the first lines of the log output with the command to the snippet we are showing in Github:

conan test conan-center-index/recipes/imake/all/test_package/conanfile.py imake/1.0.8@ --profile:host=/Users/jenkins/w/BuildSingleReference/27646/2fdb7b99-e353-499d-aa2f-153a73d71019/profile_host.txt --profile:build=default

[...]
Downloading conan_export.tgz
xorg-proto/2021.4: Downloaded recipe revision a33437da70591bb16417a77169e3f145
xorg-macros/
...

[jgsogo]

Now the actual issue, why package imake/1.0.8:c58972ce7fa79168bcca07b3885d2a61402d8743 is not found. Probably (this is just a guess) it was being uploaded when the new commit arrived and the previous job was cancelled. Sometimes the cancel request stops gracefully, sometimes it kills the job... if it was uploading the package the upload might be incomplete (some Conan files missing) and Conan thinks the package exists (IIRC it checks conaninfo.txt) but at the time it is going to download it, not all the files are available.

Why it doesn't happen more often? Because typically, when someone pushes changes to a PR and the previous job is cancelled, the changes affect the recipe-revision and it doesn't matter if one of the packages is corrupted.

I'll check in Artifactory and (probably) trigger again this PR

[madebr]

Thanks.
c3i learning new things always surprises me 😄
Great teacher you are :yoda: 🤣 😉

[jgsogo]

Thanks.
c3i learning new things always surprises me 😄
Great teacher you are :yoda: 🤣 😉

Thanks! 😊 C3i has some secrets behind the scenes... sometimes it is useful also for me to try to explain the issue, and usually, I mention @danimtb, @uilianries and @SSE4 in case it is something that helps them to improve their understanding of the system (not all of us know about every detail).

---

Indeed, that package-ID is corrupted in Artifactory, it is missing conanmanifest.txt file:

To fix it, we can just remove the package from that repository, or whatever forces Conan to build again the package and upload it. From the user's perspective, all you can do is force a new recipe revision (or a different PR number).

---

PS.- There is a feature requested to the Artifactory team to enforce consistency of packages, making the uploads transactional, or returning that a package doesn't exist if it is not valid.

[madebr]

Something happened with Windows.
Those didn't run: https://c3i.jfrog.io/c3i/misc/summary.html?json=https://c3i.jfrog.io/c3i/misc/logs/pr/6677/9-production/imake/1.0.8//summary.json

[madebr]

c3i go brrrrrrrrr

[conan-center-bot]

All green in build 12 (b7aaf7f18b60f735a154cfd19e8eeedb3998da0d):

• imake/1.0.8@:
  All packages built successfully! (All logs)

[jgsogo]

Step by step 💪

[SSE4]

paso a paso!

[madebr]

stap per stap