[Zotero 7 Beta] Problems with loading of PDF files in the Reader
The loading of PDF files has become slower and buggy in the latest beta.
Just after the text appears in the PDF Reader, it cannot be selected:
https://s3.amazonaws.com/zotero.org/images/forums/u265723/c9o4adoo7x2zomkzulmq.png
Sometimes the loading of the PDF content takes much longer than before. Even after waiting a long time, the text appears, but I still cannot do anything. I still need to wait another long time before text selection and annotations become possible.
If I try to scroll down in the PDF file during loading, at some point it suddenly goes back to the top of the PDF file.
Debug ID: D259828608
Zotero 7.0.0-beta.82+79fa160e4 (64-bit)
Windows 10
Just after the text appears in the PDF Reader, it cannot be selected:
In some cases, the right-click release to finish the text selection is not captured:[JavaScript Error: "selectionRanges[0] is undefined" {file: "resource://zotero/reader/reader.js" line: 24635}]
https://s3.amazonaws.com/zotero.org/images/forums/u265723/c9o4adoo7x2zomkzulmq.png
Sometimes the loading of the PDF content takes much longer than before. Even after waiting a long time, the text appears, but I still cannot do anything. I still need to wait another long time before text selection and annotations become possible.
If I try to scroll down in the PDF file during loading, at some point it suddenly goes back to the top of the PDF file.
Debug ID: D259828608
Zotero 7.0.0-beta.82+79fa160e4 (64-bit)
Windows 10
Sneha Ravi, A., & Dalvi, S. (2024). Liquid Marbles and Drops on Superhydrophobic Surfaces: Interfacial Aspects and Dynamics of Formation: A Review. ACS Omega, 9(11), 12307–12330. https://doi.org/10.1021/acsomega.3c07657
It seems to be related to the new figures and references extraction feature.
This is consistent with the large number of linked figures (16) and references (238) in the paper above.
It looks really nice!
https://s3.amazonaws.com/zotero.org/images/forums/u265723/86lx9ergy6ec8lp55zb1.png
But it becomes a lot slower to load.
It seems to load the pages by parts:
During loading:
https://s3.amazonaws.com/zotero.org/images/forums/u265723/1ndtrdipjbitpsvniqkn.png
After loading:
https://s3.amazonaws.com/zotero.org/images/forums/u265723/21lc6vs5ax2cgatftti0.png
Another open access example that is really slow:
Rosselló, J. M., Hoeppe, H. P., Koch, M., Lechner, C., Osterhoff, M., Vassholz, M., Hagemann, J., Möller, J., Scholz, M., Boesenberg, U., Hallmann, J., Kim, C., Zozulya, A., Lu, W., Shayduk, R., Madsen, A., Salditt, T., & Mettin, R. (2024). Jetting bubbles observed by x-ray holography at a free-electron laser: Internal structure and the effect of non-axisymmetric boundary conditions. Experiments in Fluids, 65(2), 20. https://doi.org/10.1007/s00348-023-03759-9
In the latest beta: 7.0.0-beta.83+066eda731 (64-bit).
I open this file in Troubleshooting Mode with only one tab opened: https://doi.org/10.1007/s00348-023-03759-9
I let it open for 5 minutes to make sure that there is nothing still loading or running and disable the automatic syncing.
I click and hold on the scrollbar and make it move down to the end of the paper and up to the top of the paper again 10 times quickly before releasing the scrollbar.
Then it stays in this state for more than 3 minutes:
https://s3.amazonaws.com/zotero.org/images/forums/u265723/z7aoxvfcgsdb86vd34u4.png
https://s3.amazonaws.com/zotero.org/images/forums/u265723/bd7dfcyeqmcd2wzkm421.png
Do the same in PDF.js in Firefox, it recovers after only 7 seconds.
Do the same in Adobe Acrobat Pro, it works perfectly: the content of the PDF file is always visible while scrolling down and up. Acrobat never hangs while displaying the content of the PDF file.
See this screen recording as illustration.
The only actions I am doing in there are from the mouse, and they are recorded on the right side. So nothing more than a double click to open the PDF file, scrolling, and then just wait.
The jump to the page originally displayed occurs at 0:14, ignoring the scrolling that happened in between.
I do not see the problem in PDF.js in Firefox.
You can also see a case where the text selection was successful, the highlight popup window appears briefly, and disappears (0:41).
I do not see these problems in PDF.js in Firefox: text selection is working immediately after loading.
Do you need some more examples than this one?
https://forums.zotero.org/discussion/comment/464361/#Comment_464361
For the simple test case I have described above, I get a score of 2 minutes 30 seconds on Ubuntu... so not much better than on Windows. Have you tried reproducing the detailed steps I have described (not just a simple loading)?
Maybe a screen recording will help to illustrate the comparison? You can see the comparison with PDF.js in Firefox from 1:36.
Zotero 7.0.0-beta.83+066eda731 (64-bit)
Ubuntu 24.04 LTS
Account containing only 3 items including the test case 10.1007/s00348-023-03759-9. Not synchronised.
I got a score of 2 minutes 15 seconds on that test to load the text after moving down and up 10 times.
Anyone else can test?
Debug ID: D1998651631
So hopefully someone else can bring more information on these serious loading problems...
Is there anything special about the Troubleshooting Mode that would interfere with the PDF loading process?
rendering. You wouldn’t want to use it for testing performance.
I can still trigger the problems above in normal mode, but not to a degree that is really problematic.