in Mozilla Firefox Downloads window, NVDA repeats the information associated to the file being downloaded each time the download progress is updated

[nvaccessAuto]

Reported by blindbhavya on 2014-08-25 10:10 Steps to reproduce Download something (preferably big that may take enough download time to investigate and reproduce the issue successfully). Press Ctrl + J and ensure that your focus is on the window named Libraries. Now listen to NVDA repeating what is mentioned in the summary line of this ticket. Is this an NVDA issue or an FX one? I am using FX 31

[nvaccessAuto]

Attachment 0001-In-firefox-when-the-status-of-a-download-changes-nvd.patch added by driemer.riemer@... on 2015-01-09 15:59 Description: Update: File added from Trac 0001-In-firefox-when-the-status-of-a-download-changes-nvd.patch.txt

[nvaccessAuto]

Attachment firefox.2.py added by driemer.riemer@... on 2015-03-18 16:18 Description: Update: File added from Trac

firefox.2.py.txt

[nvaccessAuto]

Attachment firefox.3.py added by driemer.riemer@... on 2015-03-19 01:35 Description: Changes in this version are: Changed if statement. Update: File added from Trac firefox.3.py.txt

[nvaccessAuto]

Attachment firefox.py added by driemer.riemer@... on 2015-03-19 02:17 Description: Most recent version as of 16. Update: File added from Trac firefox.py.txt

[nvaccessAuto]

Comment 1 by jteh on 2014-08-25 11:50 This is a tricky one. All of the information is contained in the name of the list item. The name is technically changing every time the information changes, so Firefox notifies NVDA that the name changed. In response, NVDA reads the new name. Normally, reading the new name is desirable, but here, it's obviously annoying. I guess the only way to work around this is to special case it.

[nvaccessAuto]

Comment 2 by blindbhavya on 2014-08-25 16:14 Do resolve this issue when you get the time. Thank you.

[nvaccessAuto]

Comment 3 by blindbhavya on 2014-09-15 12:41 Hi. Just posting a reminder, since quite a few new devs have come up recently. Also, do you think this ticket could be associated the keywords 'good for new devs'?

[nvaccessAuto]

Comment 4 by blindbhavya on 2014-09-15 12:45 Changes: Changed title from "in Firefox Downloads dialog NVDA keeps repeating the file being downloaded and all its other information" to "in Mozilla Firefox Downloads window, NVDA repeats the information associated to the file being downloaded each time the download progress is updated"

[nvaccessAuto]

Comment 5 by blindbhavya on 2014-09-15 12:45 Hi. I am wondering, why did you initially change the component of this ticket from Speech to Core? Please clarify m doubt.

[nvaccessAuto]

Comment 6 by jteh on 2014-09-15 21:30 Any dev is welcome to tackle any ticket, but it's probably not easy to explain how to do this one to a new developer. I'm not even 100% sure how I'd fix it myself without some thought.

It's not specifically speech related because you wouldn't fix this by modifying NVDA's speech support. You would fix it by modifying the way this event is handled for this specific case.

[nvaccessAuto]

Comment 7 by blindbhavya on 2014-09-16 01:26 Thanks for the clarification.

[nvaccessAuto]

Comment 8 by driemer.riemer@... on 2015-01-09 14:50 Hey jamie, I am gonna try to tackle this one. It may not be easy to find the accurate info, but I can probably parse the string.

[nvaccessAuto]

Comment 9 by driemer.riemer@... on 2015-01-09 15:57 Hey, Place this in your appModules folder, and we could include it into the core. Just replaceing the firefox.py would work probably. Here is both firefox.py and a patch generated off of master because i did the work in the wrong nvda repo.

Btw, would you prefer paches or linking to a git repo, or a diff?

[nvaccessAuto]

Comment 11 by mdcurran on 2015-03-18 03:56 Code review: The name of the window "library" may be a translatable string in Firefox. This will certainly work for english, but other languages may break.

I'm having trouble following your logic. to me it looks like if the new value is the came as what is cached, then you let it through to be spoken. If it is not the same, you cache it. Shouldn't it be that you let it through to be spoken only if it is different from the cache?

[nvaccessAuto]

Comment 12 by driemer.riemer@... on 2015-03-18 15:50 What I am doing is cache everything but the percentage. Then if nothing has changed ignoring the percentage, we assume we shouldn't speak the name. As with translatable strings, how do I detect the window?

[nvaccessAuto]

Comment 13 by driemer.riemer@... on 2015-03-18 16:17 Try this updated firefox.py. What I did,

*check on ROLE_LISTITEM and STATE_SELECTED

This aught to fix the issue mentioned with translatable strings.

[nvaccessAuto]

Comment 14 by jteh on 2015-03-18 22:47 You want this check to be as specific as possible. Hint: The id attribute (obj.IA2Attributes["id"]) on the list itself is "downloadsRichListBox".

[nvaccessAuto]

Comment 15 by driemer.riemer@... (in reply to comment 14) on 2015-03-19 01:38 Replying to jteh:

You want this check to be as specific as possible. Hint: The id attribute (```obj.IA2Attributes[on the list itself is "downloadsRichListBox".

Oh. I didn't know that IA2Attributes could be checked. That is a much better way of identifying the window. I also am checking on the class of the list item now.

if obj.role == controlTypes.ROLE_LISTITEM and obj.parent.IA2Attributes["id"]("id"]```)) == "downloadsRichListBox" and obj.IA2Attributes["class"] == u'download download-state': #the list's id, and the class of  the list item seem to identify this well.

The newly attached file implements these changes, and also I changed the copyright date at the top from 2012 to 2015.

[nvaccessAuto]

Comment 16 by driemer.riemer@... on 2015-03-19 02:14 Woopsi. That seems to break in the search bar.

[nvaccessAuto]

Comment 17 by briang1 on 2015-03-19 07:50 For some time now I've just set the download going, opened the library section but moved the cursor to the file below the one I'm downloaded. This seems to just give you the percentage or beeps according to what you have set up. Of course remembering to put the thing back is most important! There also does seem to be something not quite right about the options you get after a download. and I usually use the application key to get a menu to bring up the open containing folder option.

[nvaccessAuto]

Comment 18 by driemer.riemer@... on 2015-03-24 17:16 Hi, This is ready for another code review. the latest file is firefox.py.

[nvaccessAuto]

Comment 19 by driemer.riemer@... on 2015-05-19 19:31 Just a friendly reminder that this ticket is ready for a code review.

[nvaccessAuto]

Comment 21 by bdorer on 2015-05-29 16:54 The patch is working here, on NVDA next with FF40A2 so I vote for implementing.

[nvaccessAuto]

Comment 22 by driemer.riemer@... on 2015-07-09 13:24 Hi, I have another patch for firefox that I am gonna create a ticket for, but I piggy backed the code for that ontop of this ticket and thus it will be blocking this one. Is there any way we could get this into 2015.3 maybe? Thanks.

[nvaccessAuto]

Comment 23 by jteh on 2015-07-09 23:28 Is there any good reason to base the new patch on top of this one? Otherwise, this should generally be avoided.

Whether this gets into 2015.3 depends on whether I can find time to review it before then. At this stage, I'm not sure if that's going to happen.

[ehollig]

@DerekRiemer, is this something you still would like to be included into core? Should I assign you to this issue, or should I close this, and the attachment be included in a PR?

[Brian1Gaff]

This may or may not be relevant to this issue. Sometimes, when, for example downloading an update for a portable version of nvda through firefox launched by nvda, download will almost end, but not quite. apparently firefox is putting up a message saying this is not a file which is normally downloaded and wants to know if you want to finish it. This is apparently not seen by nvda, though if you suspect this is why its stopped, the context menu key does present options in the list to continue downloading. I'm not quite sure why the download window alert is not being seen by nvda. This seems to be a recent 'innovation' in Firefox 55. I imagine due to the complex file names used for snapshots its trying to be clever. If there is a way to switch this off in firefox I've not found it yet.

Brian

bglists@blueyonder.co.uk Sent via blueyonder. Please address personal email to:- briang1@blueyonder.co.uk, putting 'Brian Gaff' in the display name field.

[derekriemer]

I saw this. I don't know what this is supposed to mean, because it's just weird.

[Brian1Gaff]

I think what they are trying to do is tell people that a file is not liked for some reason, but maybe its just the complex name of snapshots that sometimes triggers the warnings. Its a shame nobody gets to hear it when as the original ticket says, all though a download all the details of that file are repeated as if its all being refreshed as the progress bar moves. The way to stop this seems to be to move away to another file, but since each firefox session in the download list only has the files in it downloaded in that session this cannot be done for the first download as you cannot cursor to an older one. Strangely in XP this is not a problem as you can see previous downloads.

Brian

bglists@blueyonder.co.uk Sent via blueyonder. Please address personal email to:- briang1@blueyonder.co.uk, putting 'Brian Gaff' in the display name field. ----- Original Message ----- From: "Derek Riemer" notifications@github.com To: "nvaccess/nvda" nvda@noreply.github.com Cc: "Brian Gaff" bglists@blueyonder.co.uk; "Comment" comment@noreply.github.com Sent: Saturday, September 02, 2017 9:29 PM Subject: Re: [nvaccess/nvda] in Mozilla Firefox Downloads window, NVDA repeats the information associated to the file being downloaded each time the download progress is updated (#4417)

I saw this. I don't know what this is supposed to mean, because it's just weird.

-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/nvaccess/nvda/issues/4417#issuecomment-326767704

[derekriemer]

No, I have seen your issue with the try build I triggered the other night. It's just strange.

[Brian1Gaff]

Maybe somebody could ask Mozilla what this new feature is all about. I've only thus far seen it on nvda downloads of snapshots. Even File Hippo where you can download huge numbers of files seems to not trigger it. Incidentally, I've read all of this thread over the years, and wondered why the author of the most successful patch did not just make it into an add-on so many could test it and see if it had any drawbacks?

Brian

bglists@blueyonder.co.uk Sent via blueyonder. Please address personal email to:- briang1@blueyonder.co.uk, putting 'Brian Gaff' in the display name field.

[derekriemer]

It's just not an issue any more unless the top file is focused.

[Brian1Gaff]

It is irritating though, if its the first download in a session as the older files are not seen in the default view in that case.

Brian

bglists@blueyonder.co.uk Sent via blueyonder. Please address personal email to:- briang1@blueyonder.co.uk, putting 'Brian Gaff' in the display name field.

[Adriani90]

@derekriemer are you planning to raise a PR with the coresponding patch? Thank you very much for your work.

[josephsl]

Hi,

Er, @JCSTeh, is this something that might be doable from Mozilla side as well? CC @MarcoZehe

[jcsteh]

I'm not sure how we could "fix" this in Firefox. The name of the list item is changing, so we fire a nameChange event accordingly. If we didn't do this, as an example, braille wouldn't update the name. A real fix for this would probably require some standardised way to indicate to a screen reader that this name shouldn't be "spoken" when it changes, but that feels pretty screen reader specific; we'd need a semantic abstraction for that.

[Adriani90]

A solution could be possible I guess if braille and speech can behave differently from an output perspective. At least this is how it works in Jaws. Braille outputs instantly and speech does it only when moving the focus away and then back on the item.