zotero-better-bibtex: Zotero broken with betterbibtex 5.1.165 and 5.1.166
Report ID: ? (Doesn’t work)
Expected behavior:
Zotero should not crash.
Actual behavior:
After Zotero updated itself to version 5.0.78, clicking any item in the library replaces the library view with the text An error has occurred. Please restart Zotero. You can report this error by selecting "Report Errors..." from the Help menu..
Disabling the better bibtex addon “solves” the problem. I tried reporting the issue using “Send Better BibTeX debug report” as stated above, but absolutely nothing happens when I select it. I presume I should have gotten an ID back.
I have however included below the error log from the “report errors” window (which is presumably what the Zotero developers use): zotero_errors.txt
- Zotero version: 5.0.78
- Better BibTex: 5.1.166
- OS: Ubuntu 16.04 x86-64
For now I will be attempting to downgrade Zotero and hope that fixes it.
About this issue
- Original URL
- State: closed
- Created 5 years ago
- Reactions: 3
- Comments: 52 (22 by maintainers)
I have it reproducible. On it.
@mgarke that’s unfortunately what happens if BBT isn’t uninstalled cleanly. With BBT running, Zotero remembers it in the column config, if the column config has unexpected items (and with BBT gone, the config for its column is unexpected), this is what Zotero does. I can’t do anything about that, sorry.
I try to do this by running my full test suite against both zotero release and zotero beta once every night. This wasn’t an incompatibility issue between them, it was actually a flaw in my test setup whose behavior leaked into the production behavior.
BBT tests starts an actual Zotero (release/beta) with an actual BBT build. On every commit I make, about 90% of my test suite is ran, excluding some tests that take a long time to complete, and only against Zotero release. Every night, and every release build, runs the full suite, against both Zotero release and beta. Incompatibilities with new versions should be caught in no more than 24 hours, and at least before I can even make a release: all releases go through my CI setup, and consequently, if even a single test fails, a new version will not be published; this is why sometimes the version skips a few, because I cannot re-release such a failed build/test.
I am deeply sorry for the mayhem I caused, but I think for a one-man spare-time project, I’ve gone the extra mile in terms of diligence and turnaround. I don’t see what I could realistically do more.
Very welcome. I don’t like to start my day off with causing such mayhem, but I’m glad I could get to it quickly with everyone helping out with logs in this thread. Super thanks people. New release is out.