0
Fixed

Bib file does not render fully when abstracts are included

f.s 1 year ago updated 1 year ago 4

I have been using BibBase for a few years and it seems there is a regression where my .bib file no longer renders completely.

Below is an example from https://bibbase.org/show?bib=fsaad.mit.edu%2Fpapers.bib&msg=embed. Only publications from 2018 to 2020 are rendered while the rest have disappeared.

If I remove all the "abstract" entries from the bib file, as in https://bibbase.org/show?bib=fsaad.mit.edu%2Fpublications.bib&msg=embed, it all renders correctly (which I am using for now on my homepage):

Any idea on why the papers do not render if I include "abstracts", and any suggestions on how to fix the issue?  Thanks.

Answer

+1
Answer
Fixed

OK, so after a lot of digging and testing with different tools I found that your server requires its requestors to send a 

'Connection': 'keep-alive'

header. Without it, it will drop the connection after some TCP packets manifesting in errors like "ConnectionResetError: [Errno 104] Connection reset by peer". So the abstracts only really matter in so much as they increased the file size beyond that limit.

I've just pushed an update to bibbase.org that makes sure that header is no included in all bibtex file requests and your original URL with the abstracts (papers.bib) is working now.

Thanks a lot for reporting. While most servers don't just hang up on requests, it is possible that some other users have been affected by this issue as well. Much appreciated!

+1

For some reason, when BibBase retrieves your bibtex file, the content abruptly ends like this:

@inproceedings{saad2018trcrpm,
author          = {Saad, Feras A. and Mansinghka, Vikash K.},
title           = {Temporally-reweighted {C}hinese restaurant process mixtures for clustering, imputing, and forecasting multivariate time series},
booktitle       = {AISTATS 2018: Proceedings of the 21st International Conference on Artificial Intelligence and Statistics},
series          = {Proceedings of Machine Learning Research},
volume          = {84},
pages           = {755--764},
publisher       = {PMLR},
address         = {Playa Blanca, Lanzarote, Canary Islands},
year            = {2018},
keywords        = {probabilistic inference, multivariate time series, nonparametric Bayes, structure learning},
url_paper       = {http://

This doesn't happen when retrieving the file via wget or curl, so I'm debugging what is causing the difference. Stay tuned.

+1
Answer
Fixed

OK, so after a lot of digging and testing with different tools I found that your server requires its requestors to send a 

'Connection': 'keep-alive'

header. Without it, it will drop the connection after some TCP packets manifesting in errors like "ConnectionResetError: [Errno 104] Connection reset by peer". So the abstracts only really matter in so much as they increased the file size beyond that limit.

I've just pushed an update to bibbase.org that makes sure that header is no included in all bibtex file requests and your original URL with the abstracts (papers.bib) is working now.

Thanks a lot for reporting. While most servers don't just hang up on requests, it is possible that some other users have been affected by this issue as well. Much appreciated!

Thank you very much for looking into this issue and quickly deploying a patch.