FS#60844 - [biber] 2.12-1 breaks biblatex
Attached to Project:
Community Packages
Opened by Simon Rose (Popkornium18) - Sunday, 18 November 2018, 00:35 GMT
Last edited by Gaetan Bisson (vesath) - Sunday, 30 December 2018, 04:29 GMT
Opened by Simon Rose (Popkornium18) - Sunday, 18 November 2018, 00:35 GMT
Last edited by Gaetan Bisson (vesath) - Sunday, 30 December 2018, 04:29 GMT
|
Details
Description:
After updating biber to version 2.12-1 my LaTeX documents won't compile anymore with the following error: ERROR - Error: Found biblatex control file version 3.4, expected version 3.5. This means that your biber (2.12) and biblatex (3.11) versions are incompatible. See compat matrix in biblatex or biber PDF documentation. Currently the only solution is to downgrade biber so that it is compatible to biblatex again. Additional info: * texlive version 2018.47465-5 Steps to reproduce: Update your system and compile a LaTeX document with biblatex and biber as the backend. |
This task depends upon
However, since it has been forty hours that this bug was reported, I'll revert biber to 2.11 myself. Cheers.
You could have give me a little bit more time.
First of all: Yes, I did a mistake and pushed it without knowing of this biblatex dependency. We should definitely add a comment in the PKGBUILD and add a rebuild.list file to the repository, because this seems to happen over and over again. I just checked the last biber bug tickets:
https://bugs.archlinux.org/task/50275
https://bugs.archlinux.org/task/60863
https://bugs.archlinux.org/task/60857
Moreover I don't think downgrading the package and using epoch was a good decision here. Why do we have no biblatex release in our repositories yet? There is a new release.. we could just rebuild, increase pkgrel of biber and have a fully working release without adding ugly epoch to the version file (sorry I don't like epoch :D ).
"Biber 2.12 is now released. It should be used in conjunction with biblatex 3.12".
Whatever, shit happens. It would be nice to have next time more time to engage. We got the ticket, don't worry. Remy wrote me even an email and was so kind to ask me before doing something (sorry remy, my bad).
Let's establish an actually truthful timeline of events.
Monday, 19 November 2018, 04:35 GMT-4 -- you, Gaetan, assign the bug to Shibumi.
Monday, 19 November 2018, 12:49 GMT-4 -- you, Gaetan, state that you were *going to give him a chance to explain himself* but won't wait any longer.
Monday, 19 November 2018, 12:56 GMT-4 -- you, Gaetan, close the bug after deciding to handle it yourself.
So, 40 hours, huh?
...
I extremely dislike this concept of cavalierly deciding to completely revert someone else's package that you don't maintain and adding an epoch in the process -- this is something that isn't even possible to take back, at all. Moreover, I seem to recall, Gaetan, that you're not fond of people changing your packages at all.
Given this, the fact that you decided to do the exact same thing to someone else, without even attempting to contact them or give them more than half a day to respond, *really* rubs me the wrong way. (Also, you're on IRC, he's on IRC, and that makes it easy to check up on each other pretty darn fast, so why not take advantage...)
Moreover, my reading of this is that you basically regard this as, like, some sort of malicious update designed to mess with people or something, since otherwise, he wouldn't need to "explain why he pushed this incompatible update" but would instead "apologize for letting a bug slip past, and cheerfully fix it". Why the focus on "explain yourself"?
Can we not have this aggressive lack of being a team? Please?
Especially when, as evidenced, the *actual* maintainer knows exactly what the problem is and has a much, much better solution. Because he's the maintainer and he tracks upstream and he's familiar with the process of tracking down domain-specific problems with his package.
Christian, I apologize for not having given you more time. My reaction was quick because I did not want to leave the packages in a broken state any longer. This bug report was created Saturday, 17 November 2018, 14:35 GMT-10. About thirty hours later I noticed the problem, found this report, and assigned it to you. About ten hours later (that is, forty hours from when the report was created) I took action. There was no good reason that I could see to leave all of our users' biblatex broken. So I have simply reverted the PKGBUILD to an earlier state, which was the least intrusive thing I could do. Yes, an epoch variable was added in the process, that's what it for; and it's just a variable, it's not "soiling" anyone's PKGBUILD...
P.S. I'd love to say that next time I'll contact you on IRC but the truth is I'll probably forget. Besides I believe we should use our bug tracker features instead of tracking down individual developers through different channels. Perhaps you can have a bot alert you on IRC when you get an email from Flyspray or more generally from an @archlinux.org address. Cheers.