Closed tris-ots closed 5 months ago
Thanks, @tris-ots. Question for @chiatt or anyone else interested:
Would it be further help for us to try to reproduce this with Arches 7.5.2, or even latest dev sources? (If you pretty much know what the problem is from looking at the report already, then further reproductions on close versions probably wouldn't add much information.)
Here's another way to reproduce (on dev/7.6.x
):
https://github.com/archesproject/arches/assets/8366985/ae149934-1af4-42ca-be5c-c0cf8feffe96
I haven't looked into the root of the problem, but I think it's safe to say published graphs should not be clonable (for now).
@chrabyrd FYI
@tris-ots can you rename this issue to something like "Cloning published resource model breaks the original and makes instances inaccessible" ?
@whatisgalen done, thanks for your experiments and for labeling this issue.
ty!
I haven't looked into the root of the problem, but I think it's safe to say published graphs should not be clonable (for now).
@whatisgalen, could you say more about that? It's not obvious to me why a graph shouldn't be cloneable just because it's published. Or does your "(for now)" parenthetical aside mean you just think they shouldn't be cloneable until this bug is fixed? (I.e., Not that there's some reason in principle why published graphs shouldn't be cloneable, just that we should prevent it until it no longer leads to a known bug.)
One thought: maybe cloning shouldn't include the published state -- you clone it, but the clone starts out as unpublished until you publish it.
/CC @chrabyrd
@kfogel apologies for the ambiguity in my comment! What I mean is that when a graph is in "published" state, if it's cloned, the clone should not also be in published state. It appears that cloning is altogether broken when done against a published state, therefore until the issue is fully resolved, a hotfix might be: clicking "clone" on a published graph should give an error and tell you to unpublish first.
If this bug is reproducible on stable/7.5.2
then we should fast-track a fix for stable/7.5.3
@kfogel apologies for the ambiguity in my comment! What I mean is that when a graph is in "published" state, if it's cloned, the clone should not also be in published state. It appears that cloning is altogether broken when done against a published state, therefore until the issue is fully resolved, a hotfix might be: clicking "clone" on a published graph should give an error and tell you to unpublish first.
Ah, okay @whatisgalen, totally understand then, and agree (unless it turns out to be just as easy to fix the underlying issue as to do the quick fix -- but I haven't looked at that part of the code so I don't know).
@kfogel I haven't had time to look into the root cause but I'd love to be surprised with an easy fix :)
@whatisgalen @kfogel Galen, thanks again for your work on this.
I've experimented with this in a 7.5.2 install and I can't seem to replicate the error where deleting the clone corrupts the original.
However, I am seeing this very generic message when I attempt to delete a cloned but unpublished RM.
Sorry! The request failed. Please try again. Contact your system administrator if the problem persists.
The install has Debug turned on, but there's nothing about this in the arches.log
.
@tris-ots @kfogel @whatisgalen
This issues has been identified and fixed in https://github.com/archesproject/arches/pull/10952
Once it's merged in it will be cherry-picked down into pre-7.6 patch releases 👍
Thank you for identifying the issue and clearly outlining reproduction steps!
@chrabyrd Thank you so much! (And +1: Hat tips to @tris-ots and @whatisgalen for their clear descriptions.)
Thanks so much @chrabyrd !
So far I've only been able to reproduce this issue in Arches 7.5.1 and with a specific ontology, and only with published RMs. I was unable to reproduce the error using 7.5.0, or with the more standard CIDOC CRM 6.2 ontology package.
Steps to reproduce:
Expected behavior:
Observed behavior:
The error: