In latest next builds, trying to update next, then makes browser events die until reboot. #7641
There's been a bug in NVDA for years that when we load source copies of NVDA, events in browsers fail to talk to us. Somehow, I always ignored this in source, because core developers experienced it too, however, with the introduction of #7563 I believe this got much worse. Every time I update next, NVDA stops responding to dynamic events until I reboot. I can still nvda+f5 to refresh however.
Since the library version is changing every time our version updates, somehow, the old code is receiving events and the new code isn't. There's something I think with this, because if I change from binary to source, and back to binary under proper working conditions, the path used to be the same, and the code would work just fine. I propose introducing a try build of NVDA that beeps upon every event responded to in a vbuff. That way, I can run that version installed, reboot, install next, then try to see if I get beeps, suggesting the old code is running.
Steps to reproduce:
Under any earlier than latest <next|master|you choose> build of NVDA
NVDA version: anything post #7563
NVDA Installed or portable: Needs to be a binary build, other than that it doesn't matter.
Does the issue still occur after restarting your PC? No
Have you tried any other versions of NVDA? I've tried source, and the several next builds.
The text was updated successfully, but these errors were encountered:
It just falls over and crashes golden cursor, see my ticket on this. A needed part of nvda seems to be missing from that snap. Brian firstname.lastname@example.org Sent via blueyonder. Please address personal email to:- email@example.com, putting 'Brian Gaff' in the display name field. ----- Original Message ----- From: "dan1982code" <firstname.lastname@example.org> To: "nvaccess/nvda" <email@example.com> Cc: "Subscribed" <firstname.lastname@example.org> Sent: Tuesday, October 03, 2017 4:47 PM Subject: Re: [nvaccess/nvda] In latest next builds, trying to update next, then makes browser events die until reboot. (#7641)…
I can confirm this, but closing Firefox completely is enough to fix it for me (reboot not required). -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: #7641 (comment)
Look at my update problems in windows 7 something is wrong with Micks idea of replacing the bits that are needed by renaming the old ones and copying new ones to run on installation or enabling of the portable that is causing things to fall over when renaming fails. This I have traced to a temp folder in the lib and lib 64 folders mostly and it means things just are not reliable any more. Also if you do an update parts of the old version now remain in the folder masking any issues that revolve around files that did not update as they still work. These various tickets I've been raising which to my mind have been ignored are at the root of the issues here and if they are not fixed when people start to install the finished update version all hell will break loose if we are not very careful. Just my opinion. Not trying to upset anyone, its just that this has been replicated on two windows 7 machines with both next and master and in both portable and installed copies, so its not finger trouble I can assure you. Its being masked as people are not sure what crud is still in the main folders of nvda. if you recall we had similar issues when the dictionary files names changed about a month back. Brian email@example.com Sent via blueyonder. Please address personal email to:- firstname.lastname@example.org, putting 'Brian Gaff' in the display name field. -----