-
Notifications
You must be signed in to change notification settings - Fork 1k
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
Tool version sorting issue: Scanpy tools default to middle-tools #15580
Comments
This is an issue with some unfortunate versioning issue with the formatting that one of us picked once ( |
I think there was some discussion as well (@nsoranzo? @bgruening?) to maybe have a smarter sort that knows how to sort |
Yes, we discussed this problem in a previous issue, but cannot find it. I'm testing my changes on that PR above to merge, hopefully that will put back the newest tool at the top. |
@mvdbeek is there a related issue or PR that closes this out in 23.0? Did I miss something? |
We can't fix this on the Galaxy side, what @pcm32 is doing in ebi-gene-expression-group/container-galaxy-sc-tertiary#266 is the way to go. |
We can't? Wasn't there some discussion that we could, by augmenting the comparison if and only if |
We could do all sorts of fancy things, but there is an established convention, there is a fix that is easy enough (ebi-gene-expression-group/container-galaxy-sc-tertiary#266), so I don't think we need to descend into something that I don't think is as easy as what you're suggesting, however if someone wants to implement this and prove that it doesn't breaking sorting for any existing tool than that would obviously be fantastic and welcome. In the meantime I don't think we should track this as an open issue. |
I neither said, nor implied, "easy" in any of my comments. Only that I remembered some discussion and I was keen to re-discover said discussion in case it had additional context I was missing. |
And I would like to close this issue as wontfix and set realistic expectations. |
Yes, that's fine, I didn't reopen or suggest that we should. Again, just trying to find the conversation I remember. |
Might be this one? #15071 |
Thank you @davelopez that looks like it! |
I mis-remembered it being focused on the general case of Wholly agree that the 2 |
@hexylena I'm happy to "hide" the tools that are causing this to bring the linage back into shape. |
Can we keep the conflicting tools visible (as they are the newest) until I push the newer version to the toolshed and that gets installed in eu? I wasn't aware that |
@pcm32 it sorts as text, you can see it on the circos tools. It's a bit unfortunate since it presents the same way. Worst for me, with the circos tools, is that it loads an older version of the tool that had some more bugs. And I'm stuck at that version since it hasn't been updated in a while. You can try |
I am planning to (find the time to) fix +galaxy10, but not the +3+galaxyX , there is already a partial fix PR from Nate #13570 . |
I have merged the tools with the version naming fix and they have been pushed to the toolshed, so now it should only be a matter for the galaxy.eu CI job that installs tools to get those updated @nomadscientist . Hopefully this should put those tools at the top. |
@pcm32 so currently today, it's still not on top |
This is the development repo for Galaxy, this is a tool issue. I can install the update on .org, only then will it appear there. |
WOOOOO! Awesome thanks! |
I did push these tools last week (Wednesday or so probably), so they should be picked up by galaxy.eu, unless there was some issue on the CI this past weekend? |
@mtekman do you happen to know why the tools are sticking to the old versions? Did you have to anchor that for a course at one point because newer tools were failing? |
ok, i'm just blind ... this is a little odd |
I checked out the source on usegalaxy.eu: the side panel is explicitly pointing at the old version, I think a restart might be enough to fix that ? ping @bgruening |
The service is restarted every day, so I doubt a restart fixes it. I can hide older tools with strange version numbers if you like. |
Could it be that the side panel has the tool version somehow hard coded in one of the files? I remember seeing this in the past somewhere. |
@pcm32 we updated over the weekend, can you please recheck? If it still does not work, please let me know which versions I should hide. |
IT WORKS! AHHHHH YAYYYY! |
Awesome, one tab less in my browser. Thanks, everyone. |
@bgruening @mvdbeek @pcm32 @hexylena |
if it's getting the wrong version in tutorial mode, then that's a different bug. we should see what's happening there. do you have a specific hands on box you can link me to? |
Scanpy PlotEmbed does this; Scanpy RunPCA, etc. etc. I think it's all the Scanpy tools. You can see it a lot here: https://training.galaxyproject.org/training-material/topics/single-cell/tutorials/scrna-case_basic-pipeline/tutorial.html |
Could you verify this against the latest versions ? I would assume that the majority of users are simply following along the instructions ... if you have more than one screen you're probably not using the overlay mode. |
we should probably add a flag where tool versions are super key, and then expose the version more visibly for those. |
Agreed - have you checked the buttons? Do you happen to know why the tutorial-mode isn't working to let us select the actual tutorial linked tools? |
@nomadscientist I'm testing this morning, it appears to be functional on EU at least. Using your two examples there, runPCA and the subsequent tool in the "Filter..." single cell tutorial, both seem to load the requested tool version |
hang onto your hat because I think Julia was an angel and updated the tool
versions in the tutorials to match
…On Fri, Jun 30, 2023 at 10:23 AM Helena ***@***.***> wrote:
@nomadscientist <https://github.com/nomadscientist> I'm testing this
morning, it appears to be functional on EU at least. Using your two
examples there, runPCA and the subsequent tool in the "Filter..." single
cell tutorial, both seem to load the requested tool version
—
Reply to this email directly, view it on GitHub
<#15580 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AKUKCSOSAF3OQKN4IHWEQS3XN2LHPANCNFSM6AAAAAAU4PZCDE>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Oh no! It is fixed! Oh boy!
…On Fri, Jun 30, 2023 at 12:23 PM Wendi Bacon ***@***.***> wrote:
hang onto your hat because I think Julia was an angel and updated the tool
versions in the tutorials to match
On Fri, Jun 30, 2023 at 10:23 AM Helena ***@***.***> wrote:
> @nomadscientist <https://github.com/nomadscientist> I'm testing this
> morning, it appears to be functional on EU at least. Using your two
> examples there, runPCA and the subsequent tool in the "Filter..." single
> cell tutorial, both seem to load the requested tool version
>
> —
> Reply to this email directly, view it on GitHub
> <#15580 (comment)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AKUKCSOSAF3OQKN4IHWEQS3XN2LHPANCNFSM6AAAAAAU4PZCDE>
> .
> You are receiving this because you were mentioned.Message ID:
> ***@***.***>
>
|
For some reason, when you select different Scanpy tools , they default to random versions in the middle of the list.
This has burned us in a few training sessions now - can someone please figure out why they aren't defaulting to the most recent?
The text was updated successfully, but these errors were encountered: