## Where communities thrive

• Join over 1.5M+ people
• Join over 100K+ communities
• Free without limits
##### Activity
• 11:14

NicolasCARPi on hypernext

try and invalidate the cache (compare)

• 10:42
NicolasCARPi commented #2936
• 09:25
MarcelBolten synchronize #2936
• 09:25

MarcelBolten on mathjax

marge mathjax-config.ts into ma… (compare)

• Sep 18 17:21
MarcelBolten commented #2936
• Sep 18 17:08
MarcelBolten commented #2936
• Sep 18 09:29
NicolasCARPi commented #2936
• Sep 18 06:39
MarcelBolten opened #2936
• Sep 18 06:33

MarcelBolten on mathjax

avoid use of type any use lazy typesetting https://do… (compare)

• Sep 17 12:48

NicolasCARPi on hypernext

hide the share button if anon u… fix the tools test (compare)

• Sep 17 08:20
yeKcim closed #21
• Sep 17 08:20
yeKcim commented #21
• Sep 17 02:10
CLAassistant commented #2935
• Sep 17 02:10
dependabot[bot] labeled #2935
• Sep 17 02:10
dependabot[bot] labeled #2935
• Sep 17 02:10
dependabot[bot] opened #2935
• Sep 17 02:10

dependabot[bot] on npm_and_yarn

Bump @types/luxon from 2.0.3 to… (compare)

• Sep 17 02:09
CLAassistant commented #2934
• Sep 17 02:09
dependabot[bot] labeled #2934
• Sep 17 02:09
dependabot[bot] labeled #2934
Nicolas CARPi
@NicolasCARPi
Frank
@FrTr
Hello @NicolasCARPi To not dicuss it on this issue elabftw/elabftw#2893 : We have the same problem. While you have it on low priority, I ask myself: How should we transfer experiments/database parts a user wants to take to another instance, where we do not have sysadmin access? The user has admin rights on both mashines. Sysadmin of the exporting machine is asking.
alexander-haller
@alexander-haller
@FrTr Hi Frank, do you have any idea how you could solve this with sysadmin access other than direkt database dumps/manipulation?
Frank
@FrTr
No, that exactly would be my question. Maybe I overlooked a feature. Maybe it would be sufficient in our case to export-dump a whole user with all items he needs, if there is a feature like this. This would make sense, due to people leaving institutions etc. that might want to take "their" data with them easily and somehow complete AND might even have a new elabFTW instance on their new place (it is getting common here!)
alexander-haller
@alexander-haller
There is no such feature afaik. I will not make such untested database operations on production. I can offer that we test this problem on my version-wise identical staging server where we can "play" all we want if you want to try something like that.
Frank
@FrTr
do you know where the problem with that elabftw/elabftw#2893 is? I mean: Could we fix the file or is there just some information not present in the dump?
alexander-haller
@alexander-haller
According to the bug "that items that have links to other items (not in the description) throw errors". And Florian says that all his experiments are linked to an element in the database.
But there seems to be the additional problem that embedded pictures get a new "hash-id" and show missing on import.
So the whole import from different servers/old versions needs some love imho.
Nicolas CARPi
@NicolasCARPi
Yeah, I think versioning should be done, so you can import a v1 archive and it will know how to deal with it differently than a v2 one.
there is already a .json file in the archive so this can be added.
alexander-haller
@alexander-haller
can/should we change the default "elabctl update" command to only pull stable versions and not betas? (with optional opt-in --beta flag)
Nicolas CARPi
@NicolasCARPi
Or optional --stable flag? :p
did you read the latest issue comment? #2919 ? The incentive is to have users actually using the beta. The vast majority of users won't update in the time window where a beta version is on master, so it's really a good way to have at least some users touching this version.
Also the beta version is called beta but in reality it could very much be a 4.1.0
Only the alpha versions, where I'm definitely not condifident about the changes will stay on next branch
Maybe elabctl update can display which version will be installed and ask for confirmation?
I understand that it makes more sense to only pull stable versions from "latest", but before updating one should have a look at the releases page and see what is "latest". And it's quite easy to pin the version in the yml file. One thing that is missing is tags like "4.1" so you can stay on 4.1 and get all patches automatically without going to 4.2 when it comes out.
Nicolas CARPi
@NicolasCARPi
alexander-haller
@alexander-haller
Read #2919 now. I completely understand your goal here. However I am inclined to think that the default of the cli should be the stable version and not require a user to read information somewhere else. I would not ask for confirmation but set an option if you want betas or not. Maybe as a tag in the yml. Other software has release channels for that (like stable/beta/alpha/daily) but this probably only works well with a really large userbase. Ideally users should have the possibility to pin version but also target stable/beta in generell.
Nicolas CARPi
@NicolasCARPi
I completely understand where you're coming from. I'll think about this, nothing is set in stone anyway :)
I think something along the lines of "A new beta version is available, would you like to try it?" could be cool
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
Hello. I'm a bit overdue to update elabftw, I'm still on 3.6.5. I want to update one thing at a time and I started with PHP (to php8). I'm getting Error with SQL Query on php8 (elabftw still 3.6.5). I was wondering, does 3.6.5 actually run on php8, or does it require php7?
Nicolas CARPi
@NicolasCARPi
no it doesn't, why are you not using Docker? It would make upgrades much easier!
Nicolas CARPi
@NicolasCARPi
If you composer install you'll see that it's php8 only, see breaking changes in release notes ;)
But really, unless it's a physical impossibility (using freebsd server), using the docker image will make your life much easier, and your instance more secure!
alexander-haller
@alexander-haller
@NicolasCARPi I think Martins case ist that he uses php8 with elabftw3.6.5 and gets an error and he is asking if "old" elabftw requires php<8.
Nicolas CARPi
@NicolasCARPi
yes, elab version <4 requires php87
sorry I mean it doesn't work with php8
so when you upgrade to php8 you have to upgrade elab to 4.0
at the risk of repeating myself, these problems are solved with containers :p
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
Thanks for the clarification! I'll look into docker.
Nicolas CARPi
@NicolasCARPi
5 replies
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
Any thoughts on this?
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
I'm going for the latter ;-)
Nicolas CARPi
@NicolasCARPi
I normally remove any "AFTER" thing because no one cares, now I'll make sure to really remove them all!!!
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
So, which line is causing the problem?
I reset the whole DB, removed line 30, and got the same error.
Nicolas CARPi
@NicolasCARPi
@saracmtd:ruhr-uni-bochum.de what's your schema version now?
I guess to find out you'll have to run queries one by one to figure out which one is causing this issue, because on the code side, I don't see where could be the issue. Did you look at this commit: elabftw/elabftw@afea3e7 ? You must not comment the entire line, as it needs to run, just change it as shown and run the file starting from that line (included).
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
AFK right now. I'll get back to you when kids are in bed 😋
Nicolas CARPi
@NicolasCARPi
:)
update kids set in_bed = 1;
but there is no where clause so all the kids in the world will go to bed :scream:
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
🤣
Martin Saraceno
@saracmtd:ruhr-uni-bochum.de
[m]
Sorry, got side-tracked... elabftw/elabftw@afea3e7 solved it! Thanks so much!
Nicolas CARPi
@NicolasCARPi
ok cool :)
alexander-haller
@alexander-haller

but there is no where clause so all the kids in the world will go to bed

coincidental: my kids went to bed last night so I guess the statement just needed 2 hours to get to my row, not bad for a global update - thanks I guess :smiley:

Nicolas CARPi
@NicolasCARPi
:D