I have been trying to repair the GitLab CI, and I'm stuck on clasp.
Whatever I try to do with clasp docker images -- and I've tried the clfoundation ones, the clasp-devel ones and even made one myself -- I get an error in the GitLab pipeline trying to start up the image. For example:
``` Using docker image sha256:5880b3b4b01af70bcf79360e28e2e9e6adbfe9599f6fc61b6815a9ebac6aea7c for rpgoldman/clasp-for-iterate:latest with digest rpgoldman/clasp-for-iterate@sha256:31d928fe1cd7e3419daf79610babe0a75d2aa7eddf34d85cc75d11831e302967 ... shell not found ```
See [this pipeline transcript](https://gitlab.common-lisp.net/rgoldman/iterate/-/jobs/42061)
This is a *stellar* error message! ;-)
Anyone know what it means? I did some googling, but no dice.
I have tried firing up this image myself and then running `make test l=clasp` in it, and the test works fine. So it's something about the interaction between the docker image and GitLab but I have no idea WTF that could be.
Any suggestions? Or do I just need to give up testing on clasp? I don't want to do that, but it seems a lot better to test iterate on n - 1 lisp implementations than to test it on zero implementations.
Thanks!
On Wed, 14 Dec 2022 at 22:10, Robert Goldman rpgoldman@sift.info wrote:
I have tried firing up this image myself and then running make test l=clasp in it, and the test works fine. So it's something about the interaction between the docker image and GitLab but I have no idea WTF that could be.
Perhaps someone from the common-lisp.net team can help?
Any suggestions? Or do I just need to give up testing on clasp? I don't want to do that, but it seems a lot better to test iterate on n - 1 lisp implementations than to test it on zero implementations.
Definitely!
Luís
On 15 Dec 2022, at 5:18, Luís Oliveira wrote:
On Wed, 14 Dec 2022 at 22:10, Robert Goldman rpgoldman@sift.info wrote:
I have tried firing up this image myself and then running make test l=clasp in it, and the test works fine. So it's something about the interaction between the docker image and GitLab but I have no idea WTF that could be.
Perhaps someone from the common-lisp.net team can help?
Any suggestions? Or do I just need to give up testing on clasp? I don't want to do that, but it seems a lot better to test iterate on n - 1 lisp implementations than to test it on zero implementations.
Definitely!hop
Luís
I'm not sure why/how, but somehow the clasp Docker images are all messed up. I think this accounts for Eric pinning the Docker image version in the ASDF tests. I grabbed that Docker image tag and put it in the Iterate tests, and now they seem fine.
So I will likely merge my branch that fixes the tests.
There are a lot of pending merge requests and issues in the ITERATE repo, but I have felt stuck because the CI was busted. I hope now we can fix some of them!