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
ENH: run: Move the image from inputs to extra_inputs #60
Conversation
We shouldn't forget to boost version dependency on datalad before releasing this one |
FWIW #3038 is merged. I thought we merged this one as well and thought to use it. |
Codecov Report
@@ Coverage Diff @@
## master #60 +/- ##
=======================================
Coverage 85.48% 85.48%
=======================================
Files 14 14
Lines 627 627
=======================================
Hits 536 536
Misses 91 91
Continue to review full report at Codecov.
|
I think I ran into this issue again, forgot that I probably installed from your PR branch, then upgraded to current master, and kaboom ... took some time to arrive back to this PR to recall details - downgraded to your PR branch and now it works again. |
I'll leave merging up to you. I'm hesitant to merge it because that means it might be released, along with the required datalad version bump, before there is a fix for datalad/datalad#3087 (fallout from datalad/datalad#3009), which breaks the "dhub://" prefix. It looks like the current plan for 3087 is to wait for revolution pr to be merged (datalad/datalad#3106), though IMO 3087 is a bigger issue than original issues addressed by 3009, and it'd be better to revert the changes from 3009 while we wait for the revolution pr. |
oh right -- good analysis, thanks! I should get back to the datalad/datalad#3106 and see what could be done, especially given @mih 's feedback. Any help would be most welcome! |
although i failed to fix that datalad/datalad#3087 yet (may be revolution would just fix it all!) is this change anyhow directly demands it? if not, I would say we should merge -- I keep needing this fix, and haven't used |
I don't have anything to add to what I said above. |
It was a useful idea to explore, but it's unlikely that anyone uses it. When an image is only available from Docker, we now support the "docker://" prefix for building Singularity images, so it's not clear that this is worth maintaining. (And, although not the adapter's fault, its existence sadly holds up the merge of dataladgh-60.) If there is something that "docker://" won't work for, we can always resurrect this.
Rebased and bumped DataLad dependency in setup.py. range-diff
|
datalad/datalad#3106 is ready for review to address the issue with adding directories. Unless some show stopped is found, it should make it into datalad 0.11.5 |
I kicked 0.11.5 out -- so after conflicts resolved we could merge/release! |
Didn't realize there was yet another round of conflicts. Will update. |
Doing so gives access to the extra_inputs argument, which isn't exposed to the user-facing interface. run_command() lacks the --rerun compatibility flag of Run.__call__, but that's not an issue because containers-run doesn't expose --rerun.
`containers-run` builds on `run`, augmenting the input list with a Singularity image. This means that if a user then puts "{inputs}" in their command, the image file is included too. This is almost certainly surprising to the user because the user has not explicitly specified the image with --input and is probably not aware that containers-run adds it. Avoid this by using run_command's recently added extra_inputs argument. Re: datalad/datalad#3038 Closes datalad#38.
extra_inputs was introduced in v0.11.2, but use v0.11.5 because that will include 514545a46 (datalad/datalad#3106) as a fix for a datalad-save regression that broke the docker adapter.
containers-run
builds onrun
, augmenting the input list with aSingularity image. This means that if a user then puts "{inputs}" in
their command, the image file is included too. This is almost
certainly surprising to the user because the user has not explicitly
specified the image with --input and is probably not aware that
containers-run adds it. Avoid this by using run_command's recently
added extra_inputs argument.
This depends on datalad/datalad#3038.
Closes #38.