I posted a comment on the "resolved" issue #3966241 on connection timeouts, please read. Thanks!
"So better servers are great, I appreciate the effort & acknowledgement of a potential problem. But this only addresses a portion of the problem. The real issue (in my mind) is that the program is incapable of recovering from connection timeouts. So, better servers are great, but it's a half-baked solution for when you have server issues, or more frequently, for those that have no choice but to work on spotty connections (i.e. college students on college wifi, anyone in America that has Comcast, etc.). I have had to work on these kinds of connections by force for school projects, and nearly stopped using writelatex altogether because I would lose hours of work in 20s increments over long projects. Note that this primarily happened to me in the spring in 2014, up to months after this issue has been marked as "completed". (I also just logged on today, a connection timeout occurred, and I lost work. I.e. any efforts to fix this in the past few months have not solved the issue.)
As a more constructive comment, I guess what I would like to see is that if you timeout on a connection, the following happens.
- restore the connection
- do NOT automatically refresh the page to the cloud version
- check to see if other contributors have edited the document since the timeout (probs need to store a most-recent-sync timestamp in the editor to make this check)
- if yes, do whatever you normally would do to merge edits (maybe that means refreshing & losing timed-out edits. I don't know.)
- if no, try to resubmit changes
- THEN refresh
(not that I know anything about how the logic works. But you guys can figure something out.)
Anyway, fixing this issue would be greatly appreciated. I have some documents that have had connections errors, let me know if sending you their links would help, or if you need anything else. Thanks for your time!"
Thanks for taking the time to provide detailed feedback. We monitor disconnections closely, and they were higher than usual late last week due to some connectivity problems at our hosting provider — sorry about this. We are working on improving our overall robustness to connection failures, as you suggest. These are some fairly complicated changes, though, so we have to make sure that we’ve carefully thought them out and tested them before release. We’ll be improving these as quickly as we can.