Topics

[nvda-devel] Screen Curtain add-on is officially legacy (end of life) #addonrelease

Lukasz Golonka
 

Hello,

On Sat, 30 Nov 2019 09:43:46 -0800
"Joseph Lee" <@joslee> wrote:

The last version of this add-on (released today) will remain on
community add-ons website under {{"legacy" section until NVDA 2019.3
stable is released, at which point it'll be removed.



This decision makes no sense in my opinion. On the one hand we are
creating section for legacy add-ons that is add-ons which are not
working with the latest NVDA version, but on the other hand, if
functionality of these add-ons is incorporated into the core you've
decided to remove them when stable version of NVDA containing this
feature comes out.
What is the difference between add-on which doesn't work with the latest version
of NVDA and add-on which is incorporated into the core, but is still
perfectly useful and working for those who cannot use latest version
for whatever reason

--
Regards
Lukasz

 

Hi,
* Add-ons not working: these are beyond author's control - NVDA made significant changes that prevent an add-on from working, even if installed on newer NVDA releases.
* Add-on features incorporated into NVDA: features from add-ons which are now included into NVDA. If authors are not careful and if said add-ons are enabled, they may introduce conflicts or unexpected behavior.

For purposes of record keeping, both types are considered legacy. This is more so now due to vast changes introduced in NVDA 2019.3 that went beyond Python 3 transition.
In case of Screen Curtain, I did say last year that the add-on was proof of concept, and once an NVDA release with screen curtain ships, it'll be discontinued. I said it that way because Leonard was working on a pull request on screen curtain, and he gave me permission to distribute his work in the form of an add-on. Also, because it was a proof of concept add-on, it didn't appear under stable add-ons section (if it did, I would have kept it past 2019.3 release for reasons you wrote below). My legacy declaration is fulfillment of this promise, hence the announcement.
Hope this clarifies things (if it didn't, let me know so I can worry about it).
Cheers,
Joseph

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 10:19 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is officially legacy (end of life) #AddonRelease

Hello,

On Sat, 30 Nov 2019 09:43:46 -0800
"Joseph Lee" <@joslee> wrote:

The last version of this add-on (released today) will remain on
community add-ons website under {{"legacy" section until NVDA 2019.3 stable is released, at which point it'll be removed.



This decision makes no sense in my opinion. On the one hand we are creating section for legacy add-ons that is add-ons which are not working with the latest NVDA version, but on the other hand, if functionality of these add-ons is incorporated into the core you've decided to remove them when stable version of NVDA containing this feature comes out.
What is the difference between add-on which doesn't work with the latest version of NVDA and add-on which is incorporated into the core, but is still perfectly useful and working for those who cannot use latest version for whatever reason

--
Regards
Lukasz

Lukasz Golonka
 

I still don't think that it should be removed from the legacy section. Even if it
is not useful for most people there still may be users who cannot migrate to 2019.3
for a number of reasons and who may wish to redownload. it. If we are going to start
removing add-ons not compatible with 2019.3 why bother to create legacy section at
all.

On Sat, 30 Nov 2019 10:43:32 -0800
"Joseph Lee" <@joslee> wrote:

Hi,
* Add-ons not working: these are beyond author's control - NVDA made significant changes that prevent an add-on from working, even if installed on newer NVDA releases.
* Add-on features incorporated into NVDA: features from add-ons which are now included into NVDA. If authors are not careful and if said add-ons are enabled, they may introduce conflicts or unexpected behavior.

For purposes of record keeping, both types are considered legacy. This is more so now due to vast changes introduced in NVDA 2019.3 that went beyond Python 3 transition.
In case of Screen Curtain, I did say last year that the add-on was proof of concept, and once an NVDA release with screen curtain ships, it'll be discontinued. I said it that way because Leonard was working on a pull request on screen curtain, and he gave me permission to distribute his work in the form of an add-on. Also, because it was a proof of concept add-on, it didn't appear under stable add-ons section (if it did, I would have kept it past 2019.3 release for reasons you wrote below). My legacy declaration is fulfillment of this promise, hence the announcement.
Hope this clarifies things (if it didn't, let me know so I can worry about it).
Cheers,
Joseph

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 10:19 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is officially legacy (end of life) #AddonRelease

Hello,

On Sat, 30 Nov 2019 09:43:46 -0800
"Joseph Lee" <@joslee> wrote:

The last version of this add-on (released today) will remain on
community add-ons website under {{"legacy" section until NVDA 2019.3 stable is released, at which point it'll be removed.



This decision makes no sense in my opinion. On the one hand we are creating section for legacy add-ons that is add-ons which are not working with the latest NVDA version, but on the other hand, if functionality of these add-ons is incorporated into the core you've decided to remove them when stable version of NVDA containing this feature comes out.
What is the difference between add-on which doesn't work with the latest version of NVDA and add-on which is incorporated into the core, but is still perfectly useful and working for those who cannot use latest version for whatever reason

--
Regards
Lukasz







 

Hi,
Hmmm, this is something we do need to think about.
I think part of the reason why people are (and would be) hesitant to upgrade to 2019.3 stable version when released is due to add-ons, which speaks volume to the work we have done so far, we are doing, and we need to do (other reasons may include compatibility, functionality, and other possibilities). To be honest with you all, I personally consider Python 3 transition work a "mixed success" - we did manage to get popular add-ons ported or got authors of these to think about what to do next, but we failed when it comes to internal coordination.
Regarding status of Screen Curtain add-on: originally, I was planning to remove the add-on completely from the website once NVDA 2019.3 shows up to everyone; I planned it as part of a general cleanup effort. Muddying the situation is the fact that this add-on's feature set is now part of an upcoming NVDA release. I may need to revisit this plan once I gather more community feedback, especially from users.
Cheers,
Joseph

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 12:20 PM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is officially legacy (end of life) #AddonRelease

I still don't think that it should be removed from the legacy section. Even if it is not useful for most people there still may be users who cannot migrate to 2019.3 for a number of reasons and who may wish to redownload. it. If we are going to start removing add-ons not compatible with 2019.3 why bother to create legacy section at all.

On Sat, 30 Nov 2019 10:43:32 -0800
"Joseph Lee" <@joslee> wrote:

Hi,
* Add-ons not working: these are beyond author's control - NVDA made significant changes that prevent an add-on from working, even if installed on newer NVDA releases.
* Add-on features incorporated into NVDA: features from add-ons which are now included into NVDA. If authors are not careful and if said add-ons are enabled, they may introduce conflicts or unexpected behavior.

For purposes of record keeping, both types are considered legacy. This is more so now due to vast changes introduced in NVDA 2019.3 that went beyond Python 3 transition.
In case of Screen Curtain, I did say last year that the add-on was proof of concept, and once an NVDA release with screen curtain ships, it'll be discontinued. I said it that way because Leonard was working on a pull request on screen curtain, and he gave me permission to distribute his work in the form of an add-on. Also, because it was a proof of concept add-on, it didn't appear under stable add-ons section (if it did, I would have kept it past 2019.3 release for reasons you wrote below). My legacy declaration is fulfillment of this promise, hence the announcement.
Hope this clarifies things (if it didn't, let me know so I can worry about it).
Cheers,
Joseph

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io
<nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 10:19 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is
officially legacy (end of life) #AddonRelease

Hello,

On Sat, 30 Nov 2019 09:43:46 -0800
"Joseph Lee" <@joslee> wrote:

The last version of this add-on (released today) will remain on
community add-ons website under {{"legacy" section until NVDA 2019.3 stable is released, at which point it'll be removed.



This decision makes no sense in my opinion. On the one hand we are creating section for legacy add-ons that is add-ons which are not working with the latest NVDA version, but on the other hand, if functionality of these add-ons is incorporated into the core you've decided to remove them when stable version of NVDA containing this feature comes out.
What is the difference between add-on which doesn't work with the
latest version of NVDA and add-on which is incorporated into the core,
but is still perfectly useful and working for those who cannot use
latest version for whatever reason

--
Regards
Lukasz







 

Well right now, its my proposal that at least for a while maybe the next little bit that we continue as we are but continue work with both versions.

The only issue with nvda 20.19.2 and others with python2 is python2 itself.

Its perfectly functional.

I am unsure about an lts release as such.

The first of these was the 2017.3 with xp, and maybe, for backward compatability there needs to be another stop point.

Either that, or maybe we should shelve nvda 2019.3 completely and put it of for 20h1 release of whatever the first 2020 release of nvda will be.

We are so close to the end of 2019, that maybe we just leave it at that right now.

Its not like its going to be a big issue with most users as it is.

With support of python2 ending, what exactly does that mean to us as users right now.

Or to put it in another way, what will eventually happen if we kept this older version.

And how long could that become a danger to us users.

I am not saying do not upgrade we must be modern and secure, but depending on how its all used, if its not going to be a problem for say the next 2 years or so, then maybe let it stew a little longer, leave it to rest maybe another 6 months maybe another year.

Eventually we will have to upgrade.

To be honest the community stuff isn't a big issue all the in house stuff would be updated quite quickly, its all the custom stuff.

I have emailed one dev of one addon I occasionally use but there are probably others.

I am not sure about maintaining 2 versions of the code base as such as thats a lot more work.

However I do think, we should at least let it rest till after christmas, heck maybe let it rest till windows 10 20h1 goes out, maybe target the big move just before that we know thats usually april/may or june, maybe march or february for the next release.

20h1 is going to be the next major release, maybe we focus the 20.1 release and skip 19.3 release or if we do it we just update with the older base.

Is it possible to keep the older libraries to run older things or is it not.

Could this be done temperarilly though.

On 1/12/2019 9:35 am, Joseph Lee wrote:
Hi,
Hmmm, this is something we do need to think about.
I think part of the reason why people are (and would be) hesitant to upgrade to 2019.3 stable version when released is due to add-ons, which speaks volume to the work we have done so far, we are doing, and we need to do (other reasons may include compatibility, functionality, and other possibilities). To be honest with you all, I personally consider Python 3 transition work a "mixed success" - we did manage to get popular add-ons ported or got authors of these to think about what to do next, but we failed when it comes to internal coordination.
Regarding status of Screen Curtain add-on: originally, I was planning to remove the add-on completely from the website once NVDA 2019.3 shows up to everyone; I planned it as part of a general cleanup effort. Muddying the situation is the fact that this add-on's feature set is now part of an upcoming NVDA release. I may need to revisit this plan once I gather more community feedback, especially from users.
Cheers,
Joseph


-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 12:20 PM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is officially legacy (end of life) #AddonRelease

I still don't think that it should be removed from the legacy section. Even if it is not useful for most people there still may be users who cannot migrate to 2019.3 for a number of reasons and who may wish to redownload. it. If we are going to start removing add-ons not compatible with 2019.3 why bother to create legacy section at all.

On Sat, 30 Nov 2019 10:43:32 -0800
"Joseph Lee" <@joslee> wrote:

Hi,
* Add-ons not working: these are beyond author's control - NVDA made significant changes that prevent an add-on from working, even if installed on newer NVDA releases.
* Add-on features incorporated into NVDA: features from add-ons which are now included into NVDA. If authors are not careful and if said add-ons are enabled, they may introduce conflicts or unexpected behavior.

For purposes of record keeping, both types are considered legacy. This is more so now due to vast changes introduced in NVDA 2019.3 that went beyond Python 3 transition.
In case of Screen Curtain, I did say last year that the add-on was proof of concept, and once an NVDA release with screen curtain ships, it'll be discontinued. I said it that way because Leonard was working on a pull request on screen curtain, and he gave me permission to distribute his work in the form of an add-on. Also, because it was a proof of concept add-on, it didn't appear under stable add-ons section (if it did, I would have kept it past 2019.3 release for reasons you wrote below). My legacy declaration is fulfillment of this promise, hence the announcement.
Hope this clarifies things (if it didn't, let me know so I can worry about it).
Cheers,
Joseph

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io
<nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 10:19 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is
officially legacy (end of life) #AddonRelease

Hello,

On Sat, 30 Nov 2019 09:43:46 -0800
"Joseph Lee" <@joslee> wrote:

The last version of this add-on (released today) will remain on
community add-ons website under {{"legacy" section until NVDA 2019.3 stable is released, at which point it'll be removed.



This decision makes no sense in my opinion. On the one hand we are creating section for legacy add-ons that is add-ons which are not working with the latest NVDA version, but on the other hand, if functionality of these add-ons is incorporated into the core you've decided to remove them when stable version of NVDA containing this feature comes out.
What is the difference between add-on which doesn't work with the
latest version of NVDA and add-on which is incorporated into the core,
but is still perfectly useful and working for those who cannot use
latest version for whatever reason

--
Regards
Lukasz










derek riemer
 



On Sat, Nov 30, 2019 at 2:39 PM Shaun Everiss <sm.everiss@...> wrote:
Well right now, its my proposal that at least for a while maybe the next
little bit that we continue as we are but continue work with both versions.
We can't maintain two parallel versions of NVDA. That would be very confusing to users, especially novice users. The transition is complete, and master can't be easily forked to bring the python3 features back to python 2. Also, our code now is set up in such a way that we are taking advantage of new python3 features.
 
The only issue with nvda 20.19.2 and others with python2 is python2 itself.

Its perfectly functional.

I don't know what that means. Can you reword that?
I am unsure about an lts release as such.
I'll bring up the possibility of that to the council, but we are trying to get other things to occur to make future addon update breakers to be easier.

The first of these was the 2017.3 with xp, and maybe, for backward
compatability there needs to be another stop point.
That wasn't an LTS as such, in that we didn't ever promise to push future changes to that release if we discovered things.

Either that, or maybe we should shelve nvda 2019.3 completely and put it
of for 20h1 release of whatever the first 2020 release of nvda will be.
We're thinking of renaming 2019.3 to 2020.1, but nothing has happened yet.

We are so close to the end of 2019, that maybe we just leave it at that
right now.

Its not like its going to be a big issue with most users as it is.

With support of python2 ending, what exactly does that mean to us as
users right now.
Nothing. The average user shouldn't even know or care what python3 is. It should be invisible. The bigger changes for the user is the new speech framework and the breaking changes that brings. The biggest problem you'll see is that addons need to be updated to a minimum tested version of 2019.3 in order for NVDA to even run them. This is because our apis broke in backwards incompatible ways, because we needed to modernize our speech infrastructure. We just decided to modernize our speech infrastructure at the same time as python3 happened, so that any compatibility issues with addons also coinsided with incompatible speech code, to minimize making users update addons twice.

Or to put it in another way, what will eventually happen if we kept this
older version.
We will have security issues, and NVDA will die because we won't be able to attract developers who want to work on crufty code that can't easily be maintained. Tech debt will accumulate, and future transitions will only become harder as we let the problem linger. Businesses won't invest in NVDA, because they would be investing in technology that is running on old and unsupported software. Remember how keysoft died before they switched to android? they couldn't push out updates because they were running on a 15 year old, unsupported windows version. It would be like that.

And how long could that b
I'm going to rephrase that as and how long until it became a danger to us users?
This depends. Not all users have the same risk profile. If the user is a human rights lawyer whose being attacked by nation state adversaries, they'd be in bad water. The typical user would probably be fine, they don't even know or care that python powers NVDA honestly. To be honest, you're on a development list, so you're seeing the natural conversation that occurs when devs need to update code. To the user, there won't be nearly as much to worry about, just run the updator, go reinstall addons. All this talk about python3 is purely playing out among developers, to hopefully give the user a smooth and seemless experience. We don't want this transition to feel like a major overhaul of NVDA.
come a danger to us users.

I am not saying do not upgrade we must be modern and secure, but
depending on how its all used, if its not going to be a problem for say
the next 2 years or so, then maybe let it stew a little longer, leave it
to rest maybe another 6 months maybe another year.
All of the code is in place now. I don't see why we need to let it delay.

Eventually we will have to upgrade.

To be honest the community stuff isn't a big issue all the in house
stuff would be updated quite quickly, its all the custom stuff.

I have emailed one dev of one addon I occasionally use but there are
probably others.
A lot of people are just not going to rework their custom stuff until they have to. Delaying this only pushes the problem further away. Most responsive developers have already updated their code. The ones who haven't, and don't before rc drops are going to have to deal with the consequences. This is kind of like the opening scene to Hitch Hikers Guide to the Galaxy, where they say "We've had the notices posted for years." In this case, however, we're doing every thing we can to promote the notices of impending incompatibility so that people can fix NVDA. Granted, if we had an addon updator already, this would go smoother.

I am not sure about maintaining 2 versions of the code base as such as thats a lot more work.
Yeah, that's not going to happen. We just don't have the resources to make that feasable, unless you want all development to crawl to a halt for however long.

However I do think, we should at least let it rest till after christmas,
heck maybe let it rest till windows 10 20h1 goes out, maybe target the
big move just before that we know thats usually april/may or june, maybe
march or february for the next release.
It probably won't happen at this point until January. I can't be certain on that, but I think beta will be at least a month, probably like 6-8 weeks.

20h1 is going to be the next major release, maybe we focus the 20.1
release and skip 19.3 release or if we do it we just update with the
older base.
Don't know if the name will be 2019.3 or 2020.1. Stay tuned.

Is it possible to keep the older libraries to run older things or is it not.

Could this be done temperarilly though.
Depends on which libraries. Likely, no.



On 1/12/2019 9:35 am, Joseph Lee wrote:
> Hi,
> Hmmm, this is something we do need to think about.
> I think part of the reason why people are (and would be) hesitant to upgrade to 2019.3 stable version when released is due to add-ons, which speaks volume to the work we have done so far, we are doing, and we need to do (other reasons may include compatibility, functionality, and other possibilities). To be honest with you all, I personally consider Python 3 transition work a "mixed success" - we did manage to get popular add-ons ported or got authors of these to think about what to do next, but we failed when it comes to internal coordination.
> Regarding status of Screen Curtain add-on: originally, I was planning to remove the add-on completely from the website once NVDA 2019.3 shows up to everyone; I planned it as part of a general cleanup effort. Muddying the situation is the fact that this add-on's feature set is now part of an upcoming NVDA release. I may need to revisit this plan once I gather more community feedback, especially from users.
> Cheers,
> Joseph
>
>
> -----Original Message-----
> From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
> Sent: Saturday, November 30, 2019 12:20 PM
> To: nvda-addons@nvda-addons.groups.io
> Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is officially legacy (end of life) #AddonRelease
>
> I still don't think that it should  be removed from the legacy  section. Even if it is not useful for most people there still may be users who cannot migrate to 2019.3 for a number of reasons and who may wish to redownload. it. If we are going to start removing add-ons not compatible with 2019.3 why bother to create legacy section at all.
>
> On Sat, 30 Nov 2019 10:43:32 -0800
> "Joseph Lee" <joseph.lee22590@...> wrote:
>
>> Hi,
>> * Add-ons not working: these are beyond author's control - NVDA made significant changes that prevent an add-on from working, even if installed on newer NVDA releases.
>> * Add-on features incorporated into NVDA: features from add-ons which are now included into NVDA. If authors are not careful and if said add-ons are enabled, they may introduce conflicts or unexpected behavior.
>>
>> For purposes of record keeping, both types are considered legacy. This is more so now due to vast changes introduced in NVDA 2019.3 that went beyond Python 3 transition.
>> In case of Screen Curtain, I did say last year that the add-on was proof of concept, and once an NVDA release with screen curtain ships, it'll be discontinued. I said it that way because Leonard was working on a pull request on screen curtain, and he gave me permission to distribute his work in the form of an add-on. Also, because it was a proof of concept add-on, it didn't appear under stable add-ons section (if it did, I would have kept it past 2019.3 release for reasons you wrote below). My legacy declaration is fulfillment of this promise, hence the announcement.
>> Hope this clarifies things (if it didn't, let me know so I can worry about it).
>> Cheers,
>> Joseph
>>
>> -----Original Message-----
>> From: nvda-addons@nvda-addons.groups.io
>> <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
>> Sent: Saturday, November 30, 2019 10:19 AM
>> To: nvda-addons@nvda-addons.groups.io
>> Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is
>> officially legacy (end of life) #AddonRelease
>>
>> Hello,
>>
>> On Sat, 30 Nov 2019 09:43:46 -0800
>> "Joseph Lee" <joseph.lee22590@...> wrote:
>>
>>> The last version of this add-on (released today) will remain on
>> community add-ons website under {{"legacy" section until NVDA 2019.3 stable is released, at which point it'll be removed.
>>
>>
>>
>> This decision makes no sense in my opinion. On the one hand we are creating section for legacy add-ons that is add-ons which are not working with the latest NVDA version, but on the other hand, if functionality of these add-ons is incorporated into the core you've decided to remove them when stable version of NVDA containing this feature comes out.
>> What is the difference between add-on which doesn't work with the
>> latest version of NVDA and add-on which is incorporated into the core,
>> but is still perfectly useful and working for those who cannot use
>> latest version for whatever reason
>>
>> --
>> Regards
>> Lukasz
>>
>>
>>
>>
>>
>>
>>
>>
>
>
>
>
>
>





--
Derek Riemer
Improving the world one byte at a time!        ⠠⠊⠍⠏⠗⠕⠧⠬ ⠮ ⠸⠺ ⠐⠕ ⠃⠽⠞⠑ ⠁⠞ ⠁ ⠐⠞⠖
•    Accessibility enthusiast.
•    Proud user of the NVDA screen reader.
•    Open source enthusiast.
•    Skier.

•    Personal website: https://derekriemer.com




 

Thanks for all that.

Well one thing we can probably agree on is release.

As a tester of things releasing things which are so close to the end of the year with all sorts of stuff going on is probably just a bit to close.

I'd actually try and release a little into next year, january or februrary or the middle of january if you had it sorted would be ok.

You are right, the user shouldn't notice anything in theory except all the new stuff.



On 1/12/2019 12:21 pm, derek riemer wrote:


On Sat, Nov 30, 2019 at 2:39 PM Shaun Everiss <sm.everiss@...> wrote:
Well right now, its my proposal that at least for a while maybe the next
little bit that we continue as we are but continue work with both versions.
We can't maintain two parallel versions of NVDA. That would be very confusing to users, especially novice users. The transition is complete, and master can't be easily forked to bring the python3 features back to python 2. Also, our code now is set up in such a way that we are taking advantage of new python3 features.
 
The only issue with nvda 20.19.2 and others with python2 is python2 itself.

Its perfectly functional.

I don't know what that means. Can you reword that?
I am unsure about an lts release as such.
I'll bring up the possibility of that to the council, but we are trying to get other things to occur to make future addon update breakers to be easier.

The first of these was the 2017.3 with xp, and maybe, for backward
compatability there needs to be another stop point.
That wasn't an LTS as such, in that we didn't ever promise to push future changes to that release if we discovered things.

Either that, or maybe we should shelve nvda 2019.3 completely and put it
of for 20h1 release of whatever the first 2020 release of nvda will be.
We're thinking of renaming 2019.3 to 2020.1, but nothing has happened yet.

We are so close to the end of 2019, that maybe we just leave it at that
right now.

Its not like its going to be a big issue with most users as it is.

With support of python2 ending, what exactly does that mean to us as
users right now.
Nothing. The average user shouldn't even know or care what python3 is. It should be invisible. The bigger changes for the user is the new speech framework and the breaking changes that brings. The biggest problem you'll see is that addons need to be updated to a minimum tested version of 2019.3 in order for NVDA to even run them. This is because our apis broke in backwards incompatible ways, because we needed to modernize our speech infrastructure. We just decided to modernize our speech infrastructure at the same time as python3 happened, so that any compatibility issues with addons also coinsided with incompatible speech code, to minimize making users update addons twice.

Or to put it in another way, what will eventually happen if we kept this
older version.
We will have security issues, and NVDA will die because we won't be able to attract developers who want to work on crufty code that can't easily be maintained. Tech debt will accumulate, and future transitions will only become harder as we let the problem linger. Businesses won't invest in NVDA, because they would be investing in technology that is running on old and unsupported software. Remember how keysoft died before they switched to android? they couldn't push out updates because they were running on a 15 year old, unsupported windows version. It would be like that.

And how long could that b
I'm going to rephrase that as and how long until it became a danger to us users?
This depends. Not all users have the same risk profile. If the user is a human rights lawyer whose being attacked by nation state adversaries, they'd be in bad water. The typical user would probably be fine, they don't even know or care that python powers NVDA honestly. To be honest, you're on a development list, so you're seeing the natural conversation that occurs when devs need to update code. To the user, there won't be nearly as much to worry about, just run the updator, go reinstall addons. All this talk about python3 is purely playing out among developers, to hopefully give the user a smooth and seemless experience. We don't want this transition to feel like a major overhaul of NVDA.
come a danger to us users.

I am not saying do not upgrade we must be modern and secure, but
depending on how its all used, if its not going to be a problem for say
the next 2 years or so, then maybe let it stew a little longer, leave it
to rest maybe another 6 months maybe another year.
All of the code is in place now. I don't see why we need to let it delay.

Eventually we will have to upgrade.

To be honest the community stuff isn't a big issue all the in house
stuff would be updated quite quickly, its all the custom stuff.

I have emailed one dev of one addon I occasionally use but there are
probably others.
A lot of people are just not going to rework their custom stuff until they have to. Delaying this only pushes the problem further away. Most responsive developers have already updated their code. The ones who haven't, and don't before rc drops are going to have to deal with the consequences. This is kind of like the opening scene to Hitch Hikers Guide to the Galaxy, where they say "We've had the notices posted for years." In this case, however, we're doing every thing we can to promote the notices of impending incompatibility so that people can fix NVDA. Granted, if we had an addon updator already, this would go smoother.

I am not sure about maintaining 2 versions of the code base as such as thats a lot more work.
Yeah, that's not going to happen. We just don't have the resources to make that feasable, unless you want all development to crawl to a halt for however long.

However I do think, we should at least let it rest till after christmas,
heck maybe let it rest till windows 10 20h1 goes out, maybe target the
big move just before that we know thats usually april/may or june, maybe
march or february for the next release.
It probably won't happen at this point until January. I can't be certain on that, but I think beta will be at least a month, probably like 6-8 weeks.

20h1 is going to be the next major release, maybe we focus the 20.1
release and skip 19.3 release or if we do it we just update with the
older base.
Don't know if the name will be 2019.3 or 2020.1. Stay tuned.

Is it possible to keep the older libraries to run older things or is it not.

Could this be done temperarilly though.
Depends on which libraries. Likely, no.



On 1/12/2019 9:35 am, Joseph Lee wrote:
> Hi,
> Hmmm, this is something we do need to think about.
> I think part of the reason why people are (and would be) hesitant to upgrade to 2019.3 stable version when released is due to add-ons, which speaks volume to the work we have done so far, we are doing, and we need to do (other reasons may include compatibility, functionality, and other possibilities). To be honest with you all, I personally consider Python 3 transition work a "mixed success" - we did manage to get popular add-ons ported or got authors of these to think about what to do next, but we failed when it comes to internal coordination.
> Regarding status of Screen Curtain add-on: originally, I was planning to remove the add-on completely from the website once NVDA 2019.3 shows up to everyone; I planned it as part of a general cleanup effort. Muddying the situation is the fact that this add-on's feature set is now part of an upcoming NVDA release. I may need to revisit this plan once I gather more community feedback, especially from users.
> Cheers,
> Joseph
>
>
> -----Original Message-----
> From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
> Sent: Saturday, November 30, 2019 12:20 PM
> To: nvda-addons@nvda-addons.groups.io
> Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is officially legacy (end of life) #AddonRelease
>
> I still don't think that it should  be removed from the legacy  section. Even if it is not useful for most people there still may be users who cannot migrate to 2019.3 for a number of reasons and who may wish to redownload. it. If we are going to start removing add-ons not compatible with 2019.3 why bother to create legacy section at all.
>
> On Sat, 30 Nov 2019 10:43:32 -0800
> "Joseph Lee" <joseph.lee22590@...> wrote:
>
>> Hi,
>> * Add-ons not working: these are beyond author's control - NVDA made significant changes that prevent an add-on from working, even if installed on newer NVDA releases.
>> * Add-on features incorporated into NVDA: features from add-ons which are now included into NVDA. If authors are not careful and if said add-ons are enabled, they may introduce conflicts or unexpected behavior.
>>
>> For purposes of record keeping, both types are considered legacy. This is more so now due to vast changes introduced in NVDA 2019.3 that went beyond Python 3 transition.
>> In case of Screen Curtain, I did say last year that the add-on was proof of concept, and once an NVDA release with screen curtain ships, it'll be discontinued. I said it that way because Leonard was working on a pull request on screen curtain, and he gave me permission to distribute his work in the form of an add-on. Also, because it was a proof of concept add-on, it didn't appear under stable add-ons section (if it did, I would have kept it past 2019.3 release for reasons you wrote below). My legacy declaration is fulfillment of this promise, hence the announcement.
>> Hope this clarifies things (if it didn't, let me know so I can worry about it).
>> Cheers,
>> Joseph
>>
>> -----Original Message-----
>> From: nvda-addons@nvda-addons.groups.io
>> <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
>> Sent: Saturday, November 30, 2019 10:19 AM
>> To: nvda-addons@nvda-addons.groups.io
>> Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is
>> officially legacy (end of life) #AddonRelease
>>
>> Hello,
>>
>> On Sat, 30 Nov 2019 09:43:46 -0800
>> "Joseph Lee" <joseph.lee22590@...> wrote:
>>
>>> The last version of this add-on (released today) will remain on
>> community add-ons website under {{"legacy" section until NVDA 2019.3 stable is released, at which point it'll be removed.
>>
>>
>>
>> This decision makes no sense in my opinion. On the one hand we are creating section for legacy add-ons that is add-ons which are not working with the latest NVDA version, but on the other hand, if functionality of these add-ons is incorporated into the core you've decided to remove them when stable version of NVDA containing this feature comes out.
>> What is the difference between add-on which doesn't work with the
>> latest version of NVDA and add-on which is incorporated into the core,
>> but is still perfectly useful and working for those who cannot use
>> latest version for whatever reason
>>
>> --
>> Regards
>> Lukasz
>>
>>
>>
>>
>>
>>
>>
>>
>
>
>
>
>
>





--
Derek Riemer
Improving the world one byte at a time!        ⠠⠊⠍⠏⠗⠕⠧⠬ ⠮ ⠸⠺ ⠐⠕ ⠃⠽⠞⠑ ⠁⠞ ⠁ ⠐⠞⠖
•    Accessibility enthusiast.
•    Proud user of the NVDA screen reader.
•    Open source enthusiast.
•    Skier.

•    Personal website: https://derekriemer.com




Kevin Cussick
 

Hi, well for me Joseph said from the start that as soon as it was in the core of Nvda it would be gone I support this.

On 30/11/2019 21:39, Shaun Everiss wrote:
Well right now, its my proposal that at least for a while maybe the next little bit that we continue as we are but continue work with both versions.
The only issue with nvda 20.19.2 and others with python2 is python2 itself.
Its perfectly functional.
I am unsure about an lts release as such.
The first of these was the 2017.3 with xp, and maybe, for backward compatability there needs to be another stop point.
Either that, or maybe we should shelve nvda 2019.3 completely and put it of for 20h1 release of whatever the first 2020 release of nvda will be.
We are so close to the end of 2019, that maybe we just leave it at that right now.
Its not like its going to be a big issue with most users as it is.
With support of python2 ending, what exactly does that mean to us as users right now.
Or to put it in another way, what will eventually happen if we kept this older version.
And how long could that become a danger to us users.
I am not saying do not upgrade we must be modern and secure, but depending on how its all used, if its not going to be a problem for say the next 2 years or so, then maybe let it stew a little longer, leave it to rest maybe another 6 months maybe another year.
Eventually we will have to upgrade.
To be honest the community stuff isn't a big issue all the in house stuff would be updated quite quickly, its all the custom stuff.
I have emailed one dev of one addon I occasionally use but there are probably others.
I am not sure about maintaining 2 versions of the code base as such as thats a lot more work.
However I do think, we should at least let it rest till after christmas, heck maybe let it rest till windows 10 20h1 goes out, maybe target the big move just before that we know thats usually april/may or june, maybe march or february for the next release.
20h1 is going to be the next major release, maybe we focus the 20.1 release and skip 19.3 release or if we do it we just update with the older base.
Is it possible to keep the older libraries to run older things or is it not.
Could this be done temperarilly though.
On 1/12/2019 9:35 am, Joseph Lee wrote:
Hi,
Hmmm, this is something we do need to think about.
I think part of the reason why people are (and would be) hesitant to upgrade to 2019.3 stable version when released is due to add-ons, which speaks volume to the work we have done so far, we are doing, and we need to do (other reasons may include compatibility, functionality, and other possibilities). To be honest with you all, I personally consider Python 3 transition work a "mixed success" - we did manage to get popular add-ons ported or got authors of these to think about what to do next, but we failed when it comes to internal coordination.
Regarding status of Screen Curtain add-on: originally, I was planning to remove the add-on completely from the website once NVDA 2019.3 shows up to everyone; I planned it as part of a general cleanup effort. Muddying the situation is the fact that this add-on's feature set is now part of an upcoming NVDA release. I may need to revisit this plan once I gather more community feedback, especially from users.
Cheers,
Joseph


-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 12:20 PM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is officially legacy (end of life) #AddonRelease

I still don't think that it should  be removed from the legacy section. Even if it is not useful for most people there still may be users who cannot migrate to 2019.3 for a number of reasons and who may wish to redownload. it. If we are going to start removing add-ons not compatible with 2019.3 why bother to create legacy section at all.

On Sat, 30 Nov 2019 10:43:32 -0800
"Joseph Lee" <@joslee> wrote:

Hi,
* Add-ons not working: these are beyond author's control - NVDA made significant changes that prevent an add-on from working, even if installed on newer NVDA releases.
* Add-on features incorporated into NVDA: features from add-ons which are now included into NVDA. If authors are not careful and if said add-ons are enabled, they may introduce conflicts or unexpected behavior.

For purposes of record keeping, both types are considered legacy. This is more so now due to vast changes introduced in NVDA 2019.3 that went beyond Python 3 transition.
In case of Screen Curtain, I did say last year that the add-on was proof of concept, and once an NVDA release with screen curtain ships, it'll be discontinued. I said it that way because Leonard was working on a pull request on screen curtain, and he gave me permission to distribute his work in the form of an add-on. Also, because it was a proof of concept add-on, it didn't appear under stable add-ons section (if it did, I would have kept it past 2019.3 release for reasons you wrote below). My legacy declaration is fulfillment of this promise, hence the announcement.
Hope this clarifies things (if it didn't, let me know so I can worry about it).
Cheers,
Joseph

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io
<nvda-addons@nvda-addons.groups.io> On Behalf Of Lukasz Golonka
Sent: Saturday, November 30, 2019 10:19 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] [nvda-devel] Screen Curtain add-on is
officially legacy (end of life) #AddonRelease

Hello,

On Sat, 30 Nov 2019 09:43:46 -0800
"Joseph Lee" <@joslee> wrote:

The last version of this add-on (released today) will remain on
community add-ons website under {{"legacy" section until NVDA 2019.3 stable is released, at which point it'll be removed.



This decision makes no sense in my opinion. On the one hand we are creating section for legacy add-ons that is add-ons which are not working with the latest NVDA version, but on the other hand, if functionality of these add-ons is incorporated into the core you've decided to remove them when stable version of NVDA containing this feature comes out.
What is the difference between add-on which doesn't work with the
latest version of NVDA and add-on which is incorporated into the core,
but is still perfectly useful and working for those who cannot use
latest version for whatever reason

--
Regards
Lukasz