Twoje komentarze

Hi Adam,

Thanks for reaching out about this! We have been trying to diagnose the slowdown all morning. It turns out that we never tested the case you describe of bulk-uploading a lot of larger files and hence didn't realize that there was a bug that hogs resources when that happens. This is what brought things to a crawl today. The system got back on its feet when I delete the files again -- I was under the impression that that would help you.

We've now also deployed a fix for the bug that caused this, so you should be able to upload the files again and this time it should be much faster and not cause any issues. It may still be safer to try and upload them in batches of 10 or so at first. Please let me know if you have any more issues.

Thanks,

Christian


Yes, it's on their end. Without them exposing this information in some way, we won't be able to display it. But you always have the option to use a shared Zotero account of which you can import the "My Library" into BibBase -- that will include pdf links. That said, bibtex files, either self-hosted, on github, or directly on BibBase, is always the most versatile option.

Thanks for changing your group settings. I was now able to find the "fulltext"s for your group library, however when I retrieve them they are just text (not PDF), e.g.:

> curl https://api.zotero.org/groups/4501563/items/KJVQ337I/fulltext | head
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0{
"content": "IFN Working Paper No. 1088, 2015\nThe National Wealth of Sweden, 1810–2014 Daniel Waldenström\nResearch Institute of Industrial Economics P.O. Box 55665\nSE-102 15 Stockholm, Sweden info@ifn.se www.ifn.se\n\nThe National Wealth of Sweden, 1810–2014*\nDaniel Waldenström† Uppsala University, IZA, CEPR and IFN\nSeptember 29, 2015\nAbstract This study presents a new database, the Swedish National Wealth Database (SNWD), which contains annual data on pr

So this doesn't really solve this problem, the PDFs added to Zotero are still not exposed by their API this way.

I've investigated this more. The fulltext feature Zotero describes in the link I posted doesn't seem to work very well. When I try it on a group with an item that has an attached PDF, I get "{}" (i.e., it doesn't find any). If I try it on your group I get:

> curl https://api.zotero.org/groups/4501563/fulltext

Forbidden

This is probably because your group is not public. We could try again after you make it "public closed", but so far this doesn't look promising.

Thanks. From the picture I was able to see that this is using a different account from the one under your name. Looking at that other account I found the problem. It was actually a bug on our end, regarding a corner-case in the zotero workflow. This is fixed now and your library now shows up:

https://bibbase.org/show?bib=https%3A%2F%2Fbibbase.org%2Fzotero%2Fmh_lenguyen&msg=embed#


(no need to use a collection)

Thanks for reporting this!

Christian

Hi Huong,

I've looked at your account but I don't see any Zotero credentials on it. What happens when you click this link: https://bibbase.org/service/zotero ? Could you send me a screenshot of what you see after you click that?

Thanks,

Christian

Url parameters, like the new noIndex one, need to come before the hash (#). They are started by a ? and separated by &. So you'll need: https://bibbase.org/service/mendeley/7a51828e-9782-3fa4-8bdb-357443954bfc/group/b49ab5c8-edb7-3544-9fc1-ebbc7f3ac750?noIndex=1

And yes, if there already are url parameters you can just append it with &noIndex=1.

And yes, to trigger an update from a new page, all you need to do is visit it in a browser. The update of the database happens asynchronously (the page itself is rendered from cache), but it should only take a few seconds to a minute depending on the data size.

Hi Adam,

Thanks for the elaborate message and attempts to debug. This helped me diagnose this pretty quickly.

Regarding the linked PDFs: You've found a bug! The issue was that we weren't yet handling filenames with spaces and special characters correctly in this context. This is fixed now and I've verified that the two papers you mentioned now download fine from the link on the page.

Regarding the per-publication pages on BibBase: yes, you are right. We deduplicate papers by their authors, title, and year and the database keeps getting updated by the last paper our servers saw with that combination. So while your other, Mendeley-based page is still live and being visited by people browsing the web, our database record for the paper keeps getting written by that old version. Once that page is removed from the web, or otherwise made inaccessible to people, the new page you are creating will be the one setting that record.

However, I recognize that this can be frustrating in some situations, like in your current migration. So we've just added a new option "noIndex=1" that you can use in the bibbase URL of your old (Mendeley-based) page to prevent it's data from being used to update the database.

Christian