Topics

Speech History 2020.1


James Scholes
 

Speech History 2020.1 is released, which is compatible with NVDA 2019.3 and 2019.3.1. It also increases the maximum stored history entries to 500. It is not compatible with any later version, such as development snapshots.

This will be pushed through Add-On Updater, or you can find it at:
https://github.com/jscholes/nvda-speech-history/releases/download/2020.1/speechHistory-2020.1.nvda-addon

Regards,

James Scholes


Cyrille
 

Hi James

Thanks for this update that was expected by many people.
For information, the description in the manifest.ini still says that this add-on allows to review the 100 last spoken texts, rather than 500.

Do you already know when you will be able to publish a version compatible with NVDA 2020.1 (now beta1 version)?
This version of the add-on is installable on NVDA 2020.1beta1 since there is no interface version upgrade. But as you told us, the add-on is not compatible with this version. Letter navigation is broken.

Thanks.
Cheers,

Cyrille

-----Message d'origine-----
De : nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> De la part de James Scholes
Envoyé : mercredi 15 avril 2020 00:58
À : nvda-addons@nvda-addons.groups.io
Objet : [nvda-addons] Speech History 2020.1

Speech History 2020.1 is released, which is compatible with NVDA 2019.3 and 2019.3.1. It also increases the maximum stored history entries to 500. It is not compatible with any later version, such as development snapshots.

This will be pushed through Add-On Updater, or you can find it at:
https://github.com/jscholes/nvda-speech-history/releases/download/2020.1/speechHistory-2020.1.nvda-addon

Regards,

James Scholes


James Scholes
 

Thanks for the hint about the manifest!

This version of the add-on is installable on NVDA 2020.1beta1 since
there is no interface version upgrade.

I don't know what you mean by "interface version upgrade". But if NVDA 2020.1 beta allows add-ons to be installed which specify their last tested version as 2019.3.1, that's a technical problem within NVDA itself.

As far as I'm aware, specifying the last tested version as 2019.3.1 is my way of saying, "this is the maximum version for which installation should be allowed". Apparently, for betas at the very least, it doesn't mean that, and I'd appreciate if someone more familiar with this area could explain why. I've already had two bug reports filed on GitHub saying it doesn't work with 2020.1, even though installing it should fail.

Do you already know when you will be able to publish a version
compatible with NVDA 2020.1

No, sorry, but it will hopefully be a lot sooner than I published this one.

Regards,

James Scholes

On 15/04/2020 at 4:38 pm, Cyrille via groups.io wrote:
Hi James
Thanks for this update that was expected by many people.
For information, the description in the manifest.ini still says that this add-on allows to review the 100 last spoken texts, rather than 500.
Do you already know when you will be able to publish a version compatible with NVDA 2020.1 (now beta1 version)?
This version of the add-on is installable on NVDA 2020.1beta1 since there is no interface version upgrade. But as you told us, the add-on is not compatible with this version. Letter navigation is broken.
Thanks.
Cheers,
Cyrille
-----Message d'origine-----
De : nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> De la part de James Scholes
Envoyé : mercredi 15 avril 2020 00:58
À : nvda-addons@nvda-addons.groups.io
Objet : [nvda-addons] Speech History 2020.1
Speech History 2020.1 is released, which is compatible with NVDA 2019.3 and 2019.3.1. It also increases the maximum stored history entries to 500. It is not compatible with any later version, such as development snapshots.
This will be pushed through Add-On Updater, or you can find it at:
https://github.com/jscholes/nvda-speech-history/releases/download/2020.1/speechHistory-2020.1.nvda-addon
Regards,
James Scholes


DaVid
 

Hi there.

The PR that I sent you a long time ago works without any aparent issues even on the NVDA's alpha versions.
Is there any reason to call the original speak function at the end of your speak function?
In my opinion You shouldn't do it because speak the message is the most important feature. An uncontrolled exception or a function that delais the call to NVDA's speak function could happen. You can deal with an exception, but not with a delay.
So, you should call the nvda's speak function first and then your code.

Regards,
David CM
Ingeniero en computación (ITCR)
Experto certificado en NVDA.
correo: dhf360@...

Encuentra mis proyectos libres en github.
y las publicaciones en mi canal de youtube.

Suscríbite y aprende en mis grupos de correo y facebook:
Programación accesible:
Envía un correo de suscripción a la dirección: programacionaccesible+subscribe@...
O participa en Programación accesible en Facebook.

tiflotronica: Si te gusta la electricidad y los chips ¡Aprende y aporta sobre este tema tan apasionante!
Envía un correo de suscripción a la dirección: tiflotronica+subscribe@...
Si te gusta la producción o ejecución de instrumentos musicales, participa en Músicos ciegos y de baja visión en facebook.
También puedes hacer tus consultas sobre tu lector de pantallas favorito en la comunidad de NVDA en español (facebook)


El mié., 15 abr. 2020 a las 15:49, James Scholes (<james@...>) escribió:
Thanks for the hint about the manifest!

 > This version of the add-on is installable on NVDA 2020.1beta1 since
there is no interface version upgrade.

I don't know what you mean by "interface version upgrade".  But if NVDA
2020.1 beta allows add-ons to be installed which specify their last
tested version as 2019.3.1, that's a technical problem within NVDA itself.

As far as I'm aware, specifying the last tested version as 2019.3.1 is
my way of saying, "this is the maximum version for which installation
should be allowed".  Apparently, for betas at the very least, it doesn't
mean that, and I'd appreciate if someone more familiar with this area
could explain why.  I've already had two bug reports filed on GitHub
saying it doesn't work with 2020.1, even though installing it should fail.

 > Do you already know when you will be able to publish a version
compatible with NVDA 2020.1

No, sorry, but it will hopefully be a lot sooner than I published this one.

Regards,

James Scholes

On 15/04/2020 at 4:38 pm, Cyrille via groups.io wrote:
> Hi James
>
> Thanks for this update that was expected by many people.
> For information, the description in the manifest.ini still says that this add-on allows to review the 100 last spoken texts, rather than 500.
>
> Do you already know when you will be able to publish a version compatible with NVDA 2020.1 (now beta1 version)?
> This version of the add-on is installable on NVDA 2020.1beta1 since there is no interface version upgrade. But as you told us, the add-on is not compatible with this version. Letter navigation is broken.
>
> Thanks.
> Cheers,
>
> Cyrille
>
>
> -----Message d'origine-----
> De : nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> De la part de James Scholes
> Envoyé : mercredi 15 avril 2020 00:58
> À : nvda-addons@nvda-addons.groups.io
> Objet : [nvda-addons] Speech History 2020.1
>
> Speech History 2020.1 is released, which is compatible with NVDA 2019.3 and 2019.3.1.  It also increases the maximum stored history entries to 500.  It is not compatible with any later version, such as development snapshots.
>
> This will be pushed through Add-On Updater, or you can find it at:
> https://github.com/jscholes/nvda-speech-history/releases/download/2020.1/speechHistory-2020.1.nvda-addon
>
> Regards,
>
> James Scholes
>
>
>
>
>
>
>




Héctor Javier Benítez Corredera
 

Hello has a pull request from David.

It explains how to solve the problem of broken characters and words.

Greetings

El 16/04/2020 a las 0:05, DaVid escribió:
Hi there.

The PR that I sent you a long time ago works without any aparent issues even on the NVDA's alpha versions.
Is there any reason to call the original speak function at the end of your speak function?
In my opinion You shouldn't do it because speak the message is the most important feature. An uncontrolled exception or a function that delais the call to NVDA's speak function could happen. You can deal with an exception, but not with a delay.
So, you should call the nvda's speak function first and then your code.

Regards,
David CM
Ingeniero en computación (ITCR)
Experto certificado en NVDA.
correo: dhf360@...

Encuentra mis proyectos libres en github.
y las publicaciones en mi canal de youtube.

Suscríbite y aprende en mis grupos de correo y facebook:
Programación accesible:
Envía un correo de suscripción a la dirección: programacionaccesible+subscribe@...
O participa en Programación accesible en Facebook.

tiflotronica: Si te gusta la electricidad y los chips ¡Aprende y aporta sobre este tema tan apasionante!
Envía un correo de suscripción a la dirección: tiflotronica+subscribe@...
Si te gusta la producción o ejecución de instrumentos musicales, participa en Músicos ciegos y de baja visión en facebook.
También puedes hacer tus consultas sobre tu lector de pantallas favorito en la comunidad de NVDA en español (facebook)


El mié., 15 abr. 2020 a las 15:49, James Scholes (<james@...>) escribió:
Thanks for the hint about the manifest!

 > This version of the add-on is installable on NVDA 2020.1beta1 since
there is no interface version upgrade.

I don't know what you mean by "interface version upgrade".  But if NVDA
2020.1 beta allows add-ons to be installed which specify their last
tested version as 2019.3.1, that's a technical problem within NVDA itself.

As far as I'm aware, specifying the last tested version as 2019.3.1 is
my way of saying, "this is the maximum version for which installation
should be allowed".  Apparently, for betas at the very least, it doesn't
mean that, and I'd appreciate if someone more familiar with this area
could explain why.  I've already had two bug reports filed on GitHub
saying it doesn't work with 2020.1, even though installing it should fail.

 > Do you already know when you will be able to publish a version
compatible with NVDA 2020.1

No, sorry, but it will hopefully be a lot sooner than I published this one.

Regards,

James Scholes

On 15/04/2020 at 4:38 pm, Cyrille via groups.io wrote:
> Hi James
>
> Thanks for this update that was expected by many people.
> For information, the description in the manifest.ini still says that this add-on allows to review the 100 last spoken texts, rather than 500.
>
> Do you already know when you will be able to publish a version compatible with NVDA 2020.1 (now beta1 version)?
> This version of the add-on is installable on NVDA 2020.1beta1 since there is no interface version upgrade. But as you told us, the add-on is not compatible with this version. Letter navigation is broken.
>
> Thanks.
> Cheers,
>
> Cyrille
>
>
> -----Message d'origine-----
> De : nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> De la part de James Scholes
> Envoyé : mercredi 15 avril 2020 00:58
> À : nvda-addons@nvda-addons.groups.io
> Objet : [nvda-addons] Speech History 2020.1
>
> Speech History 2020.1 is released, which is compatible with NVDA 2019.3 and 2019.3.1.  It also increases the maximum stored history entries to 500.  It is not compatible with any later version, such as development snapshots.
>
> This will be pushed through Add-On Updater, or you can find it at:
> https://github.com/jscholes/nvda-speech-history/releases/download/2020.1/speechHistory-2020.1.nvda-addon
>
> Regards,
>
> James Scholes
>
>
>
>
>
>
>




Tyler Spivey
 

On 4/15/2020 2:49 PM, James Scholes wrote:
As far as I'm aware, specifying the last tested version as 2019.3.1 is my way of saying, "this is the maximum version for which installation should be allowed".  Apparently, for betas at the very least, it doesn't mean that, and I'd appreciate if someone more familiar with this area could explain why.  I've already had two bug reports filed on GitHub saying it doesn't work with 2020.1, even though installing it should fail.

The reason that this breaks under 2020.1 is purely because of an incompatible change in NVDA, not the addon.
See #10879.

The question now comes up: what defines the API? Obviously not speech.speak, since it was changed and it doesn't look like it's going to be put back how it was to fix any addon which might be patching it.

I don't know what's covered by the minimum version, or what's not. What percentage of addons will need to break in order for this to be increased again, and us having another round of updating our addons?

Last tested NVDA version, from the source:
        """Tests if the addon is compatible.
        The compatibility is defined by having the required features in NVDA, and by having been tested / built against
        an API version that is still supported by this version of NVDA.
        """


Cyrille
 

Hi James,


You wrote:


I don't know what you mean by "interface version upgrade".  But if NVDA 2020.1 beta allows add-ons to be installed which specify their last tested version as 2019.3.1, that's a technical problem within NVDA itself.
As far as I'm aware, specifying the last tested version as 2019.3.1 is my way of saying, "this is the maximum version for which installation should be allowed".  Apparently, for betas at the very least, it doesn't mean that, and I'd appreciate if someone more familiar with this area could explain why.  I've already had two bug reports filed on GitHub saying it doesn't work with 2020.1, even though installing it should fail.
My answer:

By interface, I meant add-on API version, sorry.

It seems we can install add-ons specifying 2019.3.1 as last tested version on 2020.1beta1 and this probably will be the case on 2020.1 definitive release. That seems not to be a NVDA technical problem but a feature. Indeed, you can read the following lines in the file addonAPIVersion.py:

"""

This module contains add-on API version information for this build of NVDA. This file provides information on
how the API has changed as well as the range of API versions supported by this build of NVDA
"""

CURRENT = (buildVersion.version_year, buildVersion.version_major, buildVersion.version_minor)
BACK_COMPAT_TO = (2019, 3, 0)
"""
As BACK_COMPAT_TO is incremented, the changed / removed parts / or reasoning should be added below.
EG: (x, y, z): Large changes to speech.py
---
(0, 0, 0): API version zero, used to signify addons released prior to API version checks.
(2019, 3, 0): speech refactor, Python 3
"""


It seems that only changes with a big impact are listed here, probably to avoid add-on authors to publish 4 versions per year, just to update a manifest.

Cheers,

Cyrille


James Scholes
 

This makes sense; thanks for doing the research and explaining. However...

EG: (x, y, z): Large changes to speech.py
There has been tons of changes to speech.py since 2019.3.1, if I'm reading the commit history properly. So it would be nice if NVAccess followed their own advice. Admittedly I haven't had tons of time to look into this of late, so I may be wrong.

In any case, there is now a version of the add-on which works with 2020.1.

Regards,

James Scholes

On 16/04/2020 at 4:07 pm, Cyrille via groups.io wrote:
Hi James,
You wrote:

I don't know what you mean by "interface version upgrade".  But if NVDA 2020.1 beta allows add-ons to be installed which specify their last tested version as 2019.3.1, that's a technical problem within NVDA itself.
As far as I'm aware, specifying the last tested version as 2019.3.1 is my way of saying, "this is the maximum version for which installation should be allowed".  Apparently, for betas at the very least, it doesn't mean that, and I'd appreciate if someone more familiar with this area could explain why.  I've already had two bug reports filed on GitHub saying it doesn't work with 2020.1, even though installing it should fail.
My answer:
By interface, I meant add-on API version, sorry.
It seems we can install add-ons specifying 2019.3.1 as last tested version on 2020.1beta1 and this probably will be the case on 2020.1 definitive release. That seems not to be a NVDA technical problem but a feature. Indeed, you can read the following lines in the file addonAPIVersion.py:
"""
This module contains add-on API version information for this build of NVDA. This file provides information on
how the API has changed as well as the range of API versions supported by this build of NVDA
"""
CURRENT = (buildVersion.version_year, buildVersion.version_major, buildVersion.version_minor)
BACK_COMPAT_TO = (2019, 3, 0)
"""
As BACK_COMPAT_TO is incremented, the changed / removed parts / or reasoning should be added below.
EG: (x, y, z): Large changes to speech.py
---
(0, 0, 0): API version zero, used to signify addons released prior to API version checks.
(2019, 3, 0): speech refactor, Python 3
"""
It seems that only changes with a big impact are listed here, probably to avoid add-on authors to publish 4 versions per year, just to update a manifest.
Cheers,
Cyrille


Reef Turner
 

I'll be investigating #10879, this may require delaying the release until we reach a conclusion for how to proceed. Ideally this would have been picked up before Beta.

To clarify some concerns raised here, API versioning is a best effort attempt. The current stance is that any public module, class, function, or variable is included in the API and should be available and continue to provide compatible behaviour until the next API version update. This means the API is very broad, and is very hard to guarantee despite our best efforts and intentions. It should be noted however, this is more than has been offered in the past, and I expect this process to improve with time. Addon authors can help here by watching and flagging changes that may impact them, and testing their addons with alpha / beta early and often. Constructive feedback is welcome.

Kind regards,
Reef (NV Accesss)

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Tyler Spivey
Sent: Thursday, 16 April 2020 12:32 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

On 4/15/2020 2:49 PM, James Scholes wrote:
As far as I'm aware, specifying the last tested version as 2019.3.1 is
my way of saying, "this is the maximum version for which installation
should be allowed". Apparently, for betas at the very least, it
doesn't mean that, and I'd appreciate if someone more familiar with
this area could explain why. I've already had two bug reports filed
on GitHub saying it doesn't work with 2020.1, even though installing
it should fail.

The reason that this breaks under 2020.1 is purely because of an incompatible change in NVDA, not the addon.
See #10879.

The question now comes up: what defines the API? Obviously not speech.speak, since it was changed and it doesn't look like it's going to be put back how it was to fix any addon which might be patching it.

I don't know what's covered by the minimum version, or what's not. What percentage of addons will need to break in order for this to be increased again, and us having another round of updating our addons?

Last tested NVDA version, from the source:
"""Tests if the addon is compatible.
The compatibility is defined by having the required features in NVDA, and by having been tested / built against
an API version that is still supported by this version of NVDA.
"""


Reef Turner
 

Yes, there have been many changes to speech.py. The intention was to do so in a backwards compatible way as much as possible. This means ensuring that the same public API (items that do not start with an underscore) continue to exist and maintain the same behaviour.

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of James Scholes
Sent: Thursday, 16 April 2020 11:41 PM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

This makes sense; thanks for doing the research and explaining. However...

> EG: (x, y, z): Large changes to speech.py

There has been tons of changes to speech.py since 2019.3.1, if I'm reading the commit history properly. So it would be nice if NVAccess followed their own advice. Admittedly I haven't had tons of time to look into this of late, so I may be wrong.

In any case, there is now a version of the add-on which works with 2020.1.

Regards,

James Scholes

On 16/04/2020 at 4:07 pm, Cyrille via groups.io wrote:
Hi James,


You wrote:


I don't know what you mean by "interface version upgrade". But if
NVDA 2020.1 beta allows add-ons to be installed which specify their
last tested version as 2019.3.1, that's a technical problem within
NVDA itself.
As far as I'm aware, specifying the last tested version as 2019.3.1
is my way of saying, "this is the maximum version for which
installation should be allowed". Apparently, for betas at the very
least, it doesn't mean that, and I'd appreciate if someone more
familiar with this area could explain why. I've already had two bug
reports filed on GitHub saying it doesn't work with 2020.1, even
though installing it should fail.
My answer:

By interface, I meant add-on API version, sorry.

It seems we can install add-ons specifying 2019.3.1 as last tested
version on 2020.1beta1 and this probably will be the case on 2020.1
definitive release. That seems not to be a NVDA technical problem but
a feature. Indeed, you can read the following lines in the file
addonAPIVersion.py:

"""

This module contains add-on API version information for this build of
NVDA. This file provides information on how the API has changed as
well as the range of API versions supported by this build of NVDA """

CURRENT = (buildVersion.version_year, buildVersion.version_major,
buildVersion.version_minor)
BACK_COMPAT_TO = (2019, 3, 0)
"""
As BACK_COMPAT_TO is incremented, the changed / removed parts / or
reasoning should be added below.
EG: (x, y, z): Large changes to speech.py
---
(0, 0, 0): API version zero, used to signify addons released prior to
API version checks.
(2019, 3, 0): speech refactor, Python 3 """


It seems that only changes with a big impact are listed here, probably
to avoid add-on authors to publish 4 versions per year, just to update a
manifest.

Cheers,

Cyrille





James Scholes
 

All of this is fair, and thanks for providing the additional context. This add-on currently has no option other than to rely on monkey-patching, which I think probably takes all bets off the table when it comes to compatibility and any expectations. I would love to see an option for it to somehow subscribe for notifications of speech sequences in the future.

Regards,

James Scholes

On 17/04/2020 at 4:36 am, Reef Turner wrote:
I'll be investigating #10879, this may require delaying the release until we reach a conclusion for how to proceed. Ideally this would have been picked up before Beta.
To clarify some concerns raised here, API versioning is a best effort attempt. The current stance is that any public module, class, function, or variable is included in the API and should be available and continue to provide compatible behaviour until the next API version update. This means the API is very broad, and is very hard to guarantee despite our best efforts and intentions. It should be noted however, this is more than has been offered in the past, and I expect this process to improve with time. Addon authors can help here by watching and flagging changes that may impact them, and testing their addons with alpha / beta early and often. Constructive feedback is welcome.
Kind regards,
Reef (NV Accesss)
-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Tyler Spivey
Sent: Thursday, 16 April 2020 12:32 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1
On 4/15/2020 2:49 PM, James Scholes wrote:
As far as I'm aware, specifying the last tested version as 2019.3.1 is
my way of saying, "this is the maximum version for which installation
should be allowed". Apparently, for betas at the very least, it
doesn't mean that, and I'd appreciate if someone more familiar with
this area could explain why. I've already had two bug reports filed
on GitHub saying it doesn't work with 2020.1, even though installing
it should fail.

The reason that this breaks under 2020.1 is purely because of an incompatible change in NVDA, not the addon.
See #10879.
The question now comes up: what defines the API? Obviously not speech.speak, since it was changed and it doesn't look like it's going to be put back how it was to fix any addon which might be patching it.
I don't know what's covered by the minimum version, or what's not. What percentage of addons will need to break in order for this to be increased again, and us having another round of updating our addons?
Last tested NVDA version, from the source:
"""Tests if the addon is compatible.
The compatibility is defined by having the required features in NVDA, and by having been tested / built against
an API version that is still supported by this version of NVDA.
"""


 

Hi,
I believe Leonard de Ruijter or someone was working on just that - extension point action to let add-ons and other code listen to speech events without monkey patching speech functions.
Cheers,
Joseph

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of James Scholes
Sent: Friday, April 17, 2020 7:02 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

All of this is fair, and thanks for providing the additional context.
This add-on currently has no option other than to rely on monkey-patching, which I think probably takes all bets off the table when it comes to compatibility and any expectations. I would love to see an option for it to somehow subscribe for notifications of speech sequences in the future.

Regards,

James Scholes

On 17/04/2020 at 4:36 am, Reef Turner wrote:
I'll be investigating #10879, this may require delaying the release until we reach a conclusion for how to proceed. Ideally this would have been picked up before Beta.

To clarify some concerns raised here, API versioning is a best effort attempt. The current stance is that any public module, class, function, or variable is included in the API and should be available and continue to provide compatible behaviour until the next API version update. This means the API is very broad, and is very hard to guarantee despite our best efforts and intentions. It should be noted however, this is more than has been offered in the past, and I expect this process to improve with time. Addon authors can help here by watching and flagging changes that may impact them, and testing their addons with alpha / beta early and often. Constructive feedback is welcome.

Kind regards,
Reef (NV Accesss)

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io
<nvda-addons@nvda-addons.groups.io> On Behalf Of Tyler Spivey
Sent: Thursday, 16 April 2020 12:32 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

On 4/15/2020 2:49 PM, James Scholes wrote:
As far as I'm aware, specifying the last tested version as 2019.3.1
is my way of saying, "this is the maximum version for which
installation should be allowed". Apparently, for betas at the very
least, it doesn't mean that, and I'd appreciate if someone more
familiar with this area could explain why. I've already had two bug
reports filed on GitHub saying it doesn't work with 2020.1, even
though installing it should fail.

The reason that this breaks under 2020.1 is purely because of an incompatible change in NVDA, not the addon.
See #10879.

The question now comes up: what defines the API? Obviously not speech.speak, since it was changed and it doesn't look like it's going to be put back how it was to fix any addon which might be patching it.

I don't know what's covered by the minimum version, or what's not. What percentage of addons will need to break in order for this to be increased again, and us having another round of updating our addons?

Last tested NVDA version, from the source:
"""Tests if the addon is compatible.
The compatibility is defined by having the required features in NVDA, and by having been tested / built against
an API version that is still supported by this version of NVDA.
"""








Reef Turner
 

Further to this, please see PR #11040 which attempts to address the concerns here. Due to the imminent release of 2020.1, it would be great if this could be tested with the SpeechHistory addon to confirm there are no issues.

https://github.com/nvaccess/nvda/pull/11040

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Reef Turner via groups.io
Sent: Friday, 17 April 2020 11:37 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

I'll be investigating #10879, this may require delaying the release until we reach a conclusion for how to proceed. Ideally this would have been picked up before Beta.

To clarify some concerns raised here, API versioning is a best effort attempt. The current stance is that any public module, class, function, or variable is included in the API and should be available and continue to provide compatible behaviour until the next API version update. This means the API is very broad, and is very hard to guarantee despite our best efforts and intentions. It should be noted however, this is more than has been offered in the past, and I expect this process to improve with time. Addon authors can help here by watching and flagging changes that may impact them, and testing their addons with alpha / beta early and often. Constructive feedback is welcome.

Kind regards,
Reef (NV Accesss)

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Tyler Spivey
Sent: Thursday, 16 April 2020 12:32 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

On 4/15/2020 2:49 PM, James Scholes wrote:
As far as I'm aware, specifying the last tested version as 2019.3.1 is
my way of saying, "this is the maximum version for which installation
should be allowed". Apparently, for betas at the very least, it
doesn't mean that, and I'd appreciate if someone more familiar with
this area could explain why. I've already had two bug reports filed
on GitHub saying it doesn't work with 2020.1, even though installing
it should fail.

The reason that this breaks under 2020.1 is purely because of an incompatible change in NVDA, not the addon.
See #10879.

The question now comes up: what defines the API? Obviously not speech.speak, since it was changed and it doesn't look like it's going to be put back how it was to fix any addon which might be patching it.

I don't know what's covered by the minimum version, or what's not. What percentage of addons will need to break in order for this to be increased again, and us having another round of updating our addons?

Last tested NVDA version, from the source:
"""Tests if the addon is compatible.
The compatibility is defined by having the required features in NVDA, and by having been tested / built against
an API version that is still supported by this version of NVDA.
"""


James Scholes
 

Thanks, Reef. Tested with the changes from that PR and the latest version of the add-on is working fine.

Regards,

James Scholes

On 20/04/2020 at 4:09 pm, Reef Turner wrote:
Further to this, please see PR #11040 which attempts to address the concerns here. Due to the imminent release of 2020.1, it would be great if this could be tested with the SpeechHistory addon to confirm there are no issues.
https://github.com/nvaccess/nvda/pull/11040
-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Reef Turner via groups.io
Sent: Friday, 17 April 2020 11:37 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1
I'll be investigating #10879, this may require delaying the release until we reach a conclusion for how to proceed. Ideally this would have been picked up before Beta.
To clarify some concerns raised here, API versioning is a best effort attempt. The current stance is that any public module, class, function, or variable is included in the API and should be available and continue to provide compatible behaviour until the next API version update. This means the API is very broad, and is very hard to guarantee despite our best efforts and intentions. It should be noted however, this is more than has been offered in the past, and I expect this process to improve with time. Addon authors can help here by watching and flagging changes that may impact them, and testing their addons with alpha / beta early and often. Constructive feedback is welcome.
Kind regards,
Reef (NV Accesss)
-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of Tyler Spivey
Sent: Thursday, 16 April 2020 12:32 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1
On 4/15/2020 2:49 PM, James Scholes wrote:
As far as I'm aware, specifying the last tested version as 2019.3.1 is
my way of saying, "this is the maximum version for which installation
should be allowed". Apparently, for betas at the very least, it
doesn't mean that, and I'd appreciate if someone more familiar with
this area could explain why. I've already had two bug reports filed
on GitHub saying it doesn't work with 2020.1, even though installing
it should fail.

The reason that this breaks under 2020.1 is purely because of an incompatible change in NVDA, not the addon.
See #10879.
The question now comes up: what defines the API? Obviously not speech.speak, since it was changed and it doesn't look like it's going to be put back how it was to fix any addon which might be patching it.
I don't know what's covered by the minimum version, or what's not. What percentage of addons will need to break in order for this to be increased again, and us having another round of updating our addons?
Last tested NVDA version, from the source:
"""Tests if the addon is compatible.
The compatibility is defined by having the required features in NVDA, and by having been tested / built against
an API version that is still supported by this version of NVDA.
"""


Reef Turner
 

Thanks James.

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io <nvda-addons@nvda-addons.groups.io> On Behalf Of James Scholes
Sent: Tuesday, 21 April 2020 12:30 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

Thanks, Reef. Tested with the changes from that PR and the latest version of the add-on is working fine.

Regards,

James Scholes

On 20/04/2020 at 4:09 pm, Reef Turner wrote:
Further to this, please see PR #11040 which attempts to address the concerns here. Due to the imminent release of 2020.1, it would be great if this could be tested with the SpeechHistory addon to confirm there are no issues.

https://github.com/nvaccess/nvda/pull/11040

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io
<nvda-addons@nvda-addons.groups.io> On Behalf Of Reef Turner via
groups.io
Sent: Friday, 17 April 2020 11:37 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

I'll be investigating #10879, this may require delaying the release until we reach a conclusion for how to proceed. Ideally this would have been picked up before Beta.

To clarify some concerns raised here, API versioning is a best effort attempt. The current stance is that any public module, class, function, or variable is included in the API and should be available and continue to provide compatible behaviour until the next API version update. This means the API is very broad, and is very hard to guarantee despite our best efforts and intentions. It should be noted however, this is more than has been offered in the past, and I expect this process to improve with time. Addon authors can help here by watching and flagging changes that may impact them, and testing their addons with alpha / beta early and often. Constructive feedback is welcome.

Kind regards,
Reef (NV Accesss)

-----Original Message-----
From: nvda-addons@nvda-addons.groups.io
<nvda-addons@nvda-addons.groups.io> On Behalf Of Tyler Spivey
Sent: Thursday, 16 April 2020 12:32 AM
To: nvda-addons@nvda-addons.groups.io
Subject: Re: [nvda-addons] Speech History 2020.1

On 4/15/2020 2:49 PM, James Scholes wrote:
As far as I'm aware, specifying the last tested version as 2019.3.1
is my way of saying, "this is the maximum version for which
installation should be allowed". Apparently, for betas at the very
least, it doesn't mean that, and I'd appreciate if someone more
familiar with this area could explain why. I've already had two bug
reports filed on GitHub saying it doesn't work with 2020.1, even
though installing it should fail.

The reason that this breaks under 2020.1 is purely because of an incompatible change in NVDA, not the addon.
See #10879.

The question now comes up: what defines the API? Obviously not speech.speak, since it was changed and it doesn't look like it's going to be put back how it was to fix any addon which might be patching it.

I don't know what's covered by the minimum version, or what's not. What percentage of addons will need to break in order for this to be increased again, and us having another round of updating our addons?

Last tested NVDA version, from the source:
"""Tests if the addon is compatible.
The compatibility is defined by having the required features in NVDA, and by having been tested / built against
an API version that is still supported by this version of NVDA.
"""