-
Notifications
You must be signed in to change notification settings - Fork 216
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[setup-r] Ghostscript and/or qpdf installation error on Windows #812
Comments
This was an error in the chocolatey repo where we install ghostscript and qpdf from. It seems to be fixed already. [1] It has nothing to do with setup-pandoc. [1] https://github.com/r-lib/actions/actions/runs/8358418322/job/23043762630#step:4:42 |
Ah, now it is failing with
https://github.com/r-lib/callr/actions/runs/8417567882/job/23046294770#step:4:19 This sucks, because while ghostscript provides a windows installer (https://github.com/ArtifexSoftware/ghostpdl-downloads/releases), it does not work non-interactively: We could put the Ghostscript.app choco package into a GH release, and download and install it manually, I guess. It also needs that
(First the same for autohotkey.portable.) Check installation:
|
Btw. qpdf is included in Rtools43 (possible Rtools42 as well), so we don't need to download that for newer R versions. |
It seems an issue with choco service - we install R on several runners for Quarto and it only fails for some Though there is a retry mechanism and it fails all the tries.
About this, I do understand the |
Possibly, with 7zip, but the installer also creates some registry entries. So I think it is cleaner to put the choco package (and its single dependency) to GH releases and then download and install them manually with |
Yes I saw that. "Scoop" manifest did add some manual registry addition following the 10.01.0 update
It would indeed work if they maintain their choco app. It seems the issue is happening from time to time. The previously failed job are working currently |
We would download the choco package and put it on GH releases. I doubt that we would need to update it frequently. And we are already doing this, we might as well do it for qpdf. |
This should be fixed now in |
Just to share I got this error with new workflow today
I'll see if this will happen again or not. |
@cderv can you show the workflow and the workflow run? |
This was here: https://github.com/quarto-dev/quarto-cli/actions/runs/8457845577/job/23170802370?pr=9206#step:4:84 |
Thanks, for the record, there is this:
so I suspect that something happened during the ghostscript install, probably an interactive prompt. And that also made the whole step stuck:
On the one hand, this is not great. On the other hand, if it happened with the local install it can probably happen with the non-local install... In any case, definitely something to keep an eye on. |
It happened again, but only in one of the 20 windows install we do in the same CI run I don't see for now what condition or state triggers it 🤔 |
@cderv I'll see if setting the registry is important, probably not. Then we can just uncompress a ZIP and set the path instead of the choco install... |
OK, I updated |
OK, |
This issue has been automatically locked. If you believe you have found a related problem, please file a new issue and include a link to this issue |
I am trying to activate a Windows runner for data.table (Rdatatable/data.table#6022) but 2 of 3 runs I've tried so far failed at the
setup-r
phase (example failure; associated config):Poking around for how this error has been resolved elsewhere did not turn up much:
The only thing that sticks out to me as different for our setup vs. e.g. that of {lintr} which I'm more familiar with is this
setup-pandoc
step. And indeed after adding that step my pipeline passed, making me wonder if this step is recommended for Windows.This is somewhere between a bug report and a documentation request -- if adding
setup-pandoc
is really the right fix, it would be nice for it to be documented somewhere; if this apparent relationship tosetup-pandoc
is spurious, it would be nice to get a recommended fix to stabilize the windows check.The text was updated successfully, but these errors were encountered: