The VM to tag

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
17 messages Options
Reply | Threaded
Open this post in threaded view
|

The VM to tag

Eliot Miranda-2
 
Hi All,

    so it seems things are stabilizing. The CI builds.  The recent commits from Nicolas re events seem like goodness to me.  Are we agreed that I should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does anyone suggest an earlier one?  This commit has VM sources that are unchanged since Pablo Tesone's fix for gcc compilation of immediate floats. It includes the socket name lookup fixes Levente wanted, and it has very localized but useful fixes for events.  My vote would be to choose this.
_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

fniephaus
 


On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]> wrote:
 
Hi All,

    so it seems things are stabilizing. The CI builds.  The recent commits from Nicolas re events seem like goodness to me.  Are we agreed that I should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does anyone suggest an earlier one?  This commit has VM sources that are unchanged since Pablo Tesone's fix for gcc compilation of immediate floats. It includes the socket name lookup fixes Levente wanted, and it has very localized but useful fixes for events.  My vote would be to choose this.

+1

Fabio


_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

David T. Lewis
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

>  
> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave

Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave

Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave

Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 
Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
Please test it! (once the CI has built...)

Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave

Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Eliot Miranda-2
 


On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
 
Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
Please test it! (once the CI has built...)

My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:

commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
Author: Nicolas Cellier <[hidden email]>
Date:   Mon Dec 23 21:05:54 2019 +0100

    Convert charCode to latin1 encoding rather than macRoman on iOS.

    Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
    Once upon a time, the image side did use macRoman, so that explains the legacy code.
    But such legacy has become a drag, let's get rid of it!

    TODO: same changes required on legacy Mac OS
    But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.

affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m

So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.
 

Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave



--
_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Eliot Miranda-2
 
Hi Nicolas,

On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:


On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
 
Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
Please test it! (once the CI has built...)

My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:

commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
Author: Nicolas Cellier <[hidden email]>
Date:   Mon Dec 23 21:05:54 2019 +0100

    Convert charCode to latin1 encoding rather than macRoman on iOS.

    Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
    Once upon a time, the image side did use macRoman, so that explains the legacy code.
    But such legacy has become a drag, let's get rid of it!

    TODO: same changes required on legacy Mac OS
    But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.

affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m

So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.

Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
 
 

Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave



--
_,,,^..^,,,_
best, Eliot


--
_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 
Hi Eliot,
I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
Anyway, image should not use charCode, because:

    ActiveHand keyboardInterpreter

should be:

    a MacUnicodeInputInterpreter

and MacUnicodeInputInterpreter does use the utf32 code

    keyValue := evtBuf at: 6.

Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.

Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?


Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
 
Hi Nicolas,

On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:


On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
 
Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
Please test it! (once the CI has built...)

My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:

commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
Author: Nicolas Cellier <[hidden email]>
Date:   Mon Dec 23 21:05:54 2019 +0100

    Convert charCode to latin1 encoding rather than macRoman on iOS.

    Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
    Once upon a time, the image side did use macRoman, so that explains the legacy code.
    But such legacy has become a drag, let's get rid of it!

    TODO: same changes required on legacy Mac OS
    But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.

affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m

So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.

Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
 
 

Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave



--
_,,,^..^,,,_
best, Eliot


--
_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 
Hi all,
I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a

Then integrate those inbox mcz at image side:
Morphic-nice.1613
EToys-nice.374
Multilingual-nice.252

And connect smoother mouse wheel with:

     HandMorph minimumWheelDelta: 20.
     Smalltalk sendMouseWheelEvents: true.

The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
I'll commit a fix for that in inbox too.


Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
Anyway, image should not use charCode, because:

    ActiveHand keyboardInterpreter

should be:

    a MacUnicodeInputInterpreter

and MacUnicodeInputInterpreter does use the utf32 code

    keyValue := evtBuf at: 6.

Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.

Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?


Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
 
Hi Nicolas,

On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:


On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
 
Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
Please test it! (once the CI has built...)

My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:

commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
Author: Nicolas Cellier <[hidden email]>
Date:   Mon Dec 23 21:05:54 2019 +0100

    Convert charCode to latin1 encoding rather than macRoman on iOS.

    Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
    Once upon a time, the image side did use macRoman, so that explains the legacy code.
    But such legacy has become a drag, let's get rid of it!

    TODO: same changes required on legacy Mac OS
    But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.

affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m

So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.

Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
 
 

Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave



--
_,,,^..^,,,_
best, Eliot


--
_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

fniephaus
 

On Sat, 28 Dec 2019 at 4:17 pm, Nicolas Cellier <[hidden email]> wrote:
 
Hi all,
I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a

+1, but I'm for a testing period of at least two weeks. How about we integrate your image changes and once the VM changes are merged, we upgrade the VM binaries of Squeak5.3beta. Then, we start the testing period to make sure everything works fine on all platforms.

Fabio


Then integrate those inbox mcz at image side:
Morphic-nice.1613
EToys-nice.374
Multilingual-nice.252

And connect smoother mouse wheel with:

     HandMorph minimumWheelDelta: 20.
     Smalltalk sendMouseWheelEvents: true.

The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
I'll commit a fix for that in inbox too.


Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
Anyway, image should not use charCode, because:

    ActiveHand keyboardInterpreter

should be:

    a MacUnicodeInputInterpreter

and MacUnicodeInputInterpreter does use the utf32 code

    keyValue := evtBuf at: 6.

Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.

Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?


Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
 
Hi Nicolas,

On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:


On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
 
Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
Please test it! (once the CI has built...)

My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:

commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
Author: Nicolas Cellier <[hidden email]>
Date:   Mon Dec 23 21:05:54 2019 +0100

    Convert charCode to latin1 encoding rather than macRoman on iOS.

    Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
    Once upon a time, the image side did use macRoman, so that explains the legacy code.
    But such legacy has become a drag, let's get rid of it!

    TODO: same changes required on legacy Mac OS
    But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.

affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m

So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.

Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
 
 

Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave



--
_,,,^..^,,,_
best, Eliot


--
_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 


Le sam. 28 déc. 2019 à 17:04, Fabio Niephaus <[hidden email]> a écrit :
 

On Sat, 28 Dec 2019 at 4:17 pm, Nicolas Cellier <[hidden email]> wrote:
 
Hi all,
I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a

+1, but I'm for a testing period of at least two weeks. How about we integrate your image changes and once the VM changes are merged, we upgrade the VM binaries of Squeak5.3beta. Then, we start the testing period to make sure everything works fine on all platforms.

Fabio

That sounds wise.
+1


Then integrate those inbox mcz at image side:
Morphic-nice.1613
EToys-nice.374
Multilingual-nice.252

Also add Kernel-nice.1292 to the list of inbox related changes to be included.
 
And connect smoother mouse wheel with:

     HandMorph minimumWheelDelta: 20.
     Smalltalk sendMouseWheelEvents: true.

The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
I'll commit a fix for that in inbox too.


Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
Anyway, image should not use charCode, because:

    ActiveHand keyboardInterpreter

should be:

    a MacUnicodeInputInterpreter

and MacUnicodeInputInterpreter does use the utf32 code

    keyValue := evtBuf at: 6.

Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.

Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?


Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
 
Hi Nicolas,

On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:


On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
 
Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
Please test it! (once the CI has built...)

My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:

commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
Author: Nicolas Cellier <[hidden email]>
Date:   Mon Dec 23 21:05:54 2019 +0100

    Convert charCode to latin1 encoding rather than macRoman on iOS.

    Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
    Once upon a time, the image side did use macRoman, so that explains the legacy code.
    But such legacy has become a drag, let's get rid of it!

    TODO: same changes required on legacy Mac OS
    But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.

affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m

So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.

Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
 
 

Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
I can propose a fix this morning.

Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
Hi Eliot,
I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.

Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
 
On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:

> On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
> wrote:
>
> >
> > Hi All,
> >
> >     so it seems things are stabilizing. The CI builds.  The recent commits
> > from Nicolas re events seem like goodness to me.  Are we agreed that I
> > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
> > anyone suggest an earlier one?  This commit has VM sources that are
> > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
> > It includes the socket name lookup fixes Levente wanted, and it has very
> > localized but useful fixes for events.  My vote would be to choose this.
> >
>
> +1
>
> Fabio
>

+1

Dave



--
_,,,^..^,,,_
best, Eliot


--
_,,,^..^,,,_
best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

fniephaus
 
On Sat, Dec 28, 2019 at 10:20 PM Nicolas Cellier
<[hidden email]> wrote:

>
>
>
>
> Le sam. 28 déc. 2019 à 17:04, Fabio Niephaus <[hidden email]> a écrit :
>>
>>
>>
>> On Sat, 28 Dec 2019 at 4:17 pm, Nicolas Cellier <[hidden email]> wrote:
>>>
>>>
>>> Hi all,
>>> I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
>>> commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a
>>
>>
>> +1, but I'm for a testing period of at least two weeks. How about we integrate your image changes and once the VM changes are merged, we upgrade the VM binaries of Squeak5.3beta. Then, we start the testing period to make sure everything works fine on all platforms.
>>
>> Fabio
>>
> That sounds wise.
> +1
>
>>>
>>> Then integrate those inbox mcz at image side:
>>> Morphic-nice.1613
>>> EToys-nice.374
>>> Multilingual-nice.252
>>>
> Also add Kernel-nice.1292 to the list of inbox related changes to be included.

Great! Thanks a lot for cleaning up this mess, Nicolas!

>
>>>
>>> And connect smoother mouse wheel with:
>>>
>>>      HandMorph minimumWheelDelta: 20.
>>>      Smalltalk sendMouseWheelEvents: true.
>>>
>>> The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
>>> I'll commit a fix for that in inbox too.
>>>
>>>
>>> Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
>>>>
>>>> Hi Eliot,
>>>> I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
>>>> Anyway, image should not use charCode, because:
>>>>
>>>>     ActiveHand keyboardInterpreter
>>>>
>>>> should be:
>>>>
>>>>     a MacUnicodeInputInterpreter
>>>>
>>>> and MacUnicodeInputInterpreter does use the utf32 code
>>>>
>>>>     keyValue := evtBuf at: 6.
>>>>
>>>> Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
>>>> I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.
>>>>
>>>> Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
>>>> Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?
>>>>
>>>>
>>>> Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
>>>>>
>>>>>
>>>>> Hi Nicolas,
>>>>>
>>>>> On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
>>>>>>>
>>>>>>>
>>>>>>> Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
>>>>>>> Please test it! (once the CI has built...)
>>>>>>
>>>>>>
>>>>>> My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:
>>>>>>
>>>>>> commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
>>>>>> Author: Nicolas Cellier <[hidden email]>
>>>>>> Date:   Mon Dec 23 21:05:54 2019 +0100
>>>>>>
>>>>>>     Convert charCode to latin1 encoding rather than macRoman on iOS.
>>>>>>
>>>>>>     Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
>>>>>>     Once upon a time, the image side did use macRoman, so that explains the legacy code.
>>>>>>     But such legacy has become a drag, let's get rid of it!
>>>>>>
>>>>>>     TODO: same changes required on legacy Mac OS
>>>>>>     But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.
>>>>>>
>>>>>> affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m
>>>>>>
>>>>>> So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.
>>>>>
>>>>>
>>>>> Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>
>>>>>>>> I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
>>>>>>>> I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
>>>>>>>> I can propose a fix this morning.
>>>>>>>>
>>>>>>>> Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>>
>>>>>>>>> Hi Eliot,
>>>>>>>>> I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.
>>>>>>>>>
>>>>>>>>> Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:
>>>>>>>>>> >
>>>>>>>>>> > On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
>>>>>>>>>> > wrote:
>>>>>>>>>> >
>>>>>>>>>> > >
>>>>>>>>>> > > Hi All,
>>>>>>>>>> > >
>>>>>>>>>> > >     so it seems things are stabilizing. The CI builds.  The recent commits
>>>>>>>>>> > > from Nicolas re events seem like goodness to me.  Are we agreed that I
>>>>>>>>>> > > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
>>>>>>>>>> > > anyone suggest an earlier one?  This commit has VM sources that are
>>>>>>>>>> > > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
>>>>>>>>>> > > It includes the socket name lookup fixes Levente wanted, and it has very
>>>>>>>>>> > > localized but useful fixes for events.  My vote would be to choose this.
>>>>>>>>>> > >
>>>>>>>>>> >
>>>>>>>>>> > +1
>>>>>>>>>> >
>>>>>>>>>> > Fabio
>>>>>>>>>> >
>>>>>>>>>>
>>>>>>>>>> +1
>>>>>>>>>>
>>>>>>>>>> Dave
>>>>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> _,,,^..^,,,_
>>>>>> best, Eliot
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> _,,,^..^,,,_
>>>>> best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 
Update the VM to tag with b3d fix from Eliot:
a9c7fd5fc6b9c100bb5e58d09ed3c3d46f461f63

Le sam. 28 déc. 2019 à 22:22, Fabio Niephaus <[hidden email]> a écrit :
 
On Sat, Dec 28, 2019 at 10:20 PM Nicolas Cellier
<[hidden email]> wrote:
>
>
>
>
> Le sam. 28 déc. 2019 à 17:04, Fabio Niephaus <[hidden email]> a écrit :
>>
>>
>>
>> On Sat, 28 Dec 2019 at 4:17 pm, Nicolas Cellier <[hidden email]> wrote:
>>>
>>>
>>> Hi all,
>>> I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
>>> commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a
>>
>>
>> +1, but I'm for a testing period of at least two weeks. How about we integrate your image changes and once the VM changes are merged, we upgrade the VM binaries of Squeak5.3beta. Then, we start the testing period to make sure everything works fine on all platforms.
>>
>> Fabio
>>
> That sounds wise.
> +1
>
>>>
>>> Then integrate those inbox mcz at image side:
>>> Morphic-nice.1613
>>> EToys-nice.374
>>> Multilingual-nice.252
>>>
> Also add Kernel-nice.1292 to the list of inbox related changes to be included.

Great! Thanks a lot for cleaning up this mess, Nicolas!

>
>>>
>>> And connect smoother mouse wheel with:
>>>
>>>      HandMorph minimumWheelDelta: 20.
>>>      Smalltalk sendMouseWheelEvents: true.
>>>
>>> The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
>>> I'll commit a fix for that in inbox too.
>>>
>>>
>>> Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
>>>>
>>>> Hi Eliot,
>>>> I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
>>>> Anyway, image should not use charCode, because:
>>>>
>>>>     ActiveHand keyboardInterpreter
>>>>
>>>> should be:
>>>>
>>>>     a MacUnicodeInputInterpreter
>>>>
>>>> and MacUnicodeInputInterpreter does use the utf32 code
>>>>
>>>>     keyValue := evtBuf at: 6.
>>>>
>>>> Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
>>>> I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.
>>>>
>>>> Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
>>>> Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?
>>>>
>>>>
>>>> Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
>>>>>
>>>>>
>>>>> Hi Nicolas,
>>>>>
>>>>> On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
>>>>>>>
>>>>>>>
>>>>>>> Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
>>>>>>> Please test it! (once the CI has built...)
>>>>>>
>>>>>>
>>>>>> My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:
>>>>>>
>>>>>> commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
>>>>>> Author: Nicolas Cellier <[hidden email]>
>>>>>> Date:   Mon Dec 23 21:05:54 2019 +0100
>>>>>>
>>>>>>     Convert charCode to latin1 encoding rather than macRoman on iOS.
>>>>>>
>>>>>>     Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
>>>>>>     Once upon a time, the image side did use macRoman, so that explains the legacy code.
>>>>>>     But such legacy has become a drag, let's get rid of it!
>>>>>>
>>>>>>     TODO: same changes required on legacy Mac OS
>>>>>>     But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.
>>>>>>
>>>>>> affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m
>>>>>>
>>>>>> So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.
>>>>>
>>>>>
>>>>> Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>
>>>>>>>> I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
>>>>>>>> I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
>>>>>>>> I can propose a fix this morning.
>>>>>>>>
>>>>>>>> Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>>
>>>>>>>>> Hi Eliot,
>>>>>>>>> I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.
>>>>>>>>>
>>>>>>>>> Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:
>>>>>>>>>> >
>>>>>>>>>> > On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
>>>>>>>>>> > wrote:
>>>>>>>>>> >
>>>>>>>>>> > >
>>>>>>>>>> > > Hi All,
>>>>>>>>>> > >
>>>>>>>>>> > >     so it seems things are stabilizing. The CI builds.  The recent commits
>>>>>>>>>> > > from Nicolas re events seem like goodness to me.  Are we agreed that I
>>>>>>>>>> > > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
>>>>>>>>>> > > anyone suggest an earlier one?  This commit has VM sources that are
>>>>>>>>>> > > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
>>>>>>>>>> > > It includes the socket name lookup fixes Levente wanted, and it has very
>>>>>>>>>> > > localized but useful fixes for events.  My vote would be to choose this.
>>>>>>>>>> > >
>>>>>>>>>> >
>>>>>>>>>> > +1
>>>>>>>>>> >
>>>>>>>>>> > Fabio
>>>>>>>>>> >
>>>>>>>>>>
>>>>>>>>>> +1
>>>>>>>>>>
>>>>>>>>>> Dave
>>>>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> _,,,^..^,,,_
>>>>>> best, Eliot
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> _,,,^..^,,,_
>>>>> best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Eliot Miranda-2
 


On Dec 29, 2019, at 1:31 PM, Nicolas Cellier <[hidden email]> wrote:


Update the VM to tag with b3d fix from Eliot:
a9c7fd5fc6b9c100bb5e58d09ed3c3d46f461f63

+1


Le sam. 28 déc. 2019 à 22:22, Fabio Niephaus <[hidden email]> a écrit :
 
On Sat, Dec 28, 2019 at 10:20 PM Nicolas Cellier
<[hidden email]> wrote:
>
>
>
>
> Le sam. 28 déc. 2019 à 17:04, Fabio Niephaus <[hidden email]> a écrit :
>>
>>
>>
>> On Sat, 28 Dec 2019 at 4:17 pm, Nicolas Cellier <[hidden email]> wrote:
>>>
>>>
>>> Hi all,
>>> I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
>>> commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a
>>
>>
>> +1, but I'm for a testing period of at least two weeks. How about we integrate your image changes and once the VM changes are merged, we upgrade the VM binaries of Squeak5.3beta. Then, we start the testing period to make sure everything works fine on all platforms.
>>
>> Fabio
>>
> That sounds wise.
> +1
>
>>>
>>> Then integrate those inbox mcz at image side:
>>> Morphic-nice.1613
>>> EToys-nice.374
>>> Multilingual-nice.252
>>>
> Also add Kernel-nice.1292 to the list of inbox related changes to be included.

Great! Thanks a lot for cleaning up this mess, Nicolas!

>
>>>
>>> And connect smoother mouse wheel with:
>>>
>>>      HandMorph minimumWheelDelta: 20.
>>>      Smalltalk sendMouseWheelEvents: true.
>>>
>>> The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
>>> I'll commit a fix for that in inbox too.
>>>
>>>
>>> Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
>>>>
>>>> Hi Eliot,
>>>> I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
>>>> Anyway, image should not use charCode, because:
>>>>
>>>>     ActiveHand keyboardInterpreter
>>>>
>>>> should be:
>>>>
>>>>     a MacUnicodeInputInterpreter
>>>>
>>>> and MacUnicodeInputInterpreter does use the utf32 code
>>>>
>>>>     keyValue := evtBuf at: 6.
>>>>
>>>> Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
>>>> I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.
>>>>
>>>> Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
>>>> Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?
>>>>
>>>>
>>>> Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
>>>>>
>>>>>
>>>>> Hi Nicolas,
>>>>>
>>>>> On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
>>>>>>>
>>>>>>>
>>>>>>> Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
>>>>>>> Please test it! (once the CI has built...)
>>>>>>
>>>>>>
>>>>>> My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:
>>>>>>
>>>>>> commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
>>>>>> Author: Nicolas Cellier <[hidden email]>
>>>>>> Date:   Mon Dec 23 21:05:54 2019 +0100
>>>>>>
>>>>>>     Convert charCode to latin1 encoding rather than macRoman on iOS.
>>>>>>
>>>>>>     Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
>>>>>>     Once upon a time, the image side did use macRoman, so that explains the legacy code.
>>>>>>     But such legacy has become a drag, let's get rid of it!
>>>>>>
>>>>>>     TODO: same changes required on legacy Mac OS
>>>>>>     But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.
>>>>>>
>>>>>> affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m
>>>>>>
>>>>>> So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.
>>>>>
>>>>>
>>>>> Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>
>>>>>>>> I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
>>>>>>>> I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
>>>>>>>> I can propose a fix this morning.
>>>>>>>>
>>>>>>>> Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>>
>>>>>>>>> Hi Eliot,
>>>>>>>>> I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.
>>>>>>>>>
>>>>>>>>> Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:
>>>>>>>>>> >
>>>>>>>>>> > On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
>>>>>>>>>> > wrote:
>>>>>>>>>> >
>>>>>>>>>> > >
>>>>>>>>>> > > Hi All,
>>>>>>>>>> > >
>>>>>>>>>> > >     so it seems things are stabilizing. The CI builds.  The recent commits
>>>>>>>>>> > > from Nicolas re events seem like goodness to me.  Are we agreed that I
>>>>>>>>>> > > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
>>>>>>>>>> > > anyone suggest an earlier one?  This commit has VM sources that are
>>>>>>>>>> > > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
>>>>>>>>>> > > It includes the socket name lookup fixes Levente wanted, and it has very
>>>>>>>>>> > > localized but useful fixes for events.  My vote would be to choose this.
>>>>>>>>>> > >
>>>>>>>>>> >
>>>>>>>>>> > +1
>>>>>>>>>> >
>>>>>>>>>> > Fabio
>>>>>>>>>> >
>>>>>>>>>>
>>>>>>>>>> +1
>>>>>>>>>>
>>>>>>>>>> Dave
>>>>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> _,,,^..^,,,_
>>>>>> best, Eliot
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> _,,,^..^,,,_
>>>>> best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Nicolas Cellier
 
Probably last message of the year from this European time zone:

please tag SHA f219b7218fad8f083c28f645d2e976f9cb826344
it fixes horizontal mouse (trackpad) wheel on OSX.

With this, we have same conventions on 3 OS, OSX, linux, windows
at least when emulated on same physical machine/trackpad.

Le lun. 30 déc. 2019 à 01:21, Eliot Miranda <[hidden email]> a écrit :
 


On Dec 29, 2019, at 1:31 PM, Nicolas Cellier <[hidden email]> wrote:


Update the VM to tag with b3d fix from Eliot:
a9c7fd5fc6b9c100bb5e58d09ed3c3d46f461f63

+1


Le sam. 28 déc. 2019 à 22:22, Fabio Niephaus <[hidden email]> a écrit :
 
On Sat, Dec 28, 2019 at 10:20 PM Nicolas Cellier
<[hidden email]> wrote:
>
>
>
>
> Le sam. 28 déc. 2019 à 17:04, Fabio Niephaus <[hidden email]> a écrit :
>>
>>
>>
>> On Sat, 28 Dec 2019 at 4:17 pm, Nicolas Cellier <[hidden email]> wrote:
>>>
>>>
>>> Hi all,
>>> I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
>>> commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a
>>
>>
>> +1, but I'm for a testing period of at least two weeks. How about we integrate your image changes and once the VM changes are merged, we upgrade the VM binaries of Squeak5.3beta. Then, we start the testing period to make sure everything works fine on all platforms.
>>
>> Fabio
>>
> That sounds wise.
> +1
>
>>>
>>> Then integrate those inbox mcz at image side:
>>> Morphic-nice.1613
>>> EToys-nice.374
>>> Multilingual-nice.252
>>>
> Also add Kernel-nice.1292 to the list of inbox related changes to be included.

Great! Thanks a lot for cleaning up this mess, Nicolas!

>
>>>
>>> And connect smoother mouse wheel with:
>>>
>>>      HandMorph minimumWheelDelta: 20.
>>>      Smalltalk sendMouseWheelEvents: true.
>>>
>>> The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
>>> I'll commit a fix for that in inbox too.
>>>
>>>
>>> Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
>>>>
>>>> Hi Eliot,
>>>> I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
>>>> Anyway, image should not use charCode, because:
>>>>
>>>>     ActiveHand keyboardInterpreter
>>>>
>>>> should be:
>>>>
>>>>     a MacUnicodeInputInterpreter
>>>>
>>>> and MacUnicodeInputInterpreter does use the utf32 code
>>>>
>>>>     keyValue := evtBuf at: 6.
>>>>
>>>> Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
>>>> I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.
>>>>
>>>> Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
>>>> Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?
>>>>
>>>>
>>>> Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
>>>>>
>>>>>
>>>>> Hi Nicolas,
>>>>>
>>>>> On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
>>>>>>>
>>>>>>>
>>>>>>> Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
>>>>>>> Please test it! (once the CI has built...)
>>>>>>
>>>>>>
>>>>>> My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:
>>>>>>
>>>>>> commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
>>>>>> Author: Nicolas Cellier <[hidden email]>
>>>>>> Date:   Mon Dec 23 21:05:54 2019 +0100
>>>>>>
>>>>>>     Convert charCode to latin1 encoding rather than macRoman on iOS.
>>>>>>
>>>>>>     Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
>>>>>>     Once upon a time, the image side did use macRoman, so that explains the legacy code.
>>>>>>     But such legacy has become a drag, let's get rid of it!
>>>>>>
>>>>>>     TODO: same changes required on legacy Mac OS
>>>>>>     But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.
>>>>>>
>>>>>> affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m
>>>>>>
>>>>>> So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.
>>>>>
>>>>>
>>>>> Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>
>>>>>>>> I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
>>>>>>>> I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
>>>>>>>> I can propose a fix this morning.
>>>>>>>>
>>>>>>>> Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>>
>>>>>>>>> Hi Eliot,
>>>>>>>>> I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.
>>>>>>>>>
>>>>>>>>> Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:
>>>>>>>>>> >
>>>>>>>>>> > On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
>>>>>>>>>> > wrote:
>>>>>>>>>> >
>>>>>>>>>> > >
>>>>>>>>>> > > Hi All,
>>>>>>>>>> > >
>>>>>>>>>> > >     so it seems things are stabilizing. The CI builds.  The recent commits
>>>>>>>>>> > > from Nicolas re events seem like goodness to me.  Are we agreed that I
>>>>>>>>>> > > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
>>>>>>>>>> > > anyone suggest an earlier one?  This commit has VM sources that are
>>>>>>>>>> > > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
>>>>>>>>>> > > It includes the socket name lookup fixes Levente wanted, and it has very
>>>>>>>>>> > > localized but useful fixes for events.  My vote would be to choose this.
>>>>>>>>>> > >
>>>>>>>>>> >
>>>>>>>>>> > +1
>>>>>>>>>> >
>>>>>>>>>> > Fabio
>>>>>>>>>> >
>>>>>>>>>>
>>>>>>>>>> +1
>>>>>>>>>>
>>>>>>>>>> Dave
>>>>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> _,,,^..^,,,_
>>>>>> best, Eliot
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> _,,,^..^,,,_
>>>>> best, Eliot
Reply | Threaded
Open this post in threaded view
|

Re: The VM to tag

Eliot Miranda-2
 


On Dec 31, 2019, at 2:30 PM, Nicolas Cellier <[hidden email]> wrote:


Probably last message of the year from this European time zone:

please tag SHA f219b7218fad8f083c28f645d2e976f9cb826344
it fixes horizontal mouse (trackpad) wheel on OSX.

With this, we have same conventions on 3 OS, OSX, linux, windows
at least when emulated on same physical machine/trackpad.

<3  _/\_

Happy new year to everybody!!


Le lun. 30 déc. 2019 à 01:21, Eliot Miranda <[hidden email]> a écrit :
 


On Dec 29, 2019, at 1:31 PM, Nicolas Cellier <[hidden email]> wrote:


Update the VM to tag with b3d fix from Eliot:
a9c7fd5fc6b9c100bb5e58d09ed3c3d46f461f63

+1


Le sam. 28 déc. 2019 à 22:22, Fabio Niephaus <[hidden email]> a écrit :
 
On Sat, Dec 28, 2019 at 10:20 PM Nicolas Cellier
<[hidden email]> wrote:
>
>
>
>
> Le sam. 28 déc. 2019 à 17:04, Fabio Niephaus <[hidden email]> a écrit :
>>
>>
>>
>> On Sat, 28 Dec 2019 at 4:17 pm, Nicolas Cellier <[hidden email]> wrote:
>>>
>>>
>>> Hi all,
>>> I'd prefer to have the mouse wheel fix integrated (PR #461 and PR #462), that is:
>>> commit SHA: 83e43ee7138f8a1b76ed774c3972f2691776778a
>>
>>
>> +1, but I'm for a testing period of at least two weeks. How about we integrate your image changes and once the VM changes are merged, we upgrade the VM binaries of Squeak5.3beta. Then, we start the testing period to make sure everything works fine on all platforms.
>>
>> Fabio
>>
> That sounds wise.
> +1
>
>>>
>>> Then integrate those inbox mcz at image side:
>>> Morphic-nice.1613
>>> EToys-nice.374
>>> Multilingual-nice.252
>>>
> Also add Kernel-nice.1292 to the list of inbox related changes to be included.

Great! Thanks a lot for cleaning up this mess, Nicolas!

>
>>>
>>> And connect smoother mouse wheel with:
>>>
>>>      HandMorph minimumWheelDelta: 20.
>>>      Smalltalk sendMouseWheelEvents: true.
>>>
>>> The cmd+key problem is at image side, we have to fix recent changes from Marcel now that linux (x11) ctrl+a produces ASCII control again.
>>> I'll commit a fix for that in inbox too.
>>>
>>>
>>> Le ven. 27 déc. 2019 à 14:01, Nicolas Cellier <[hidden email]> a écrit :
>>>>
>>>> Hi Eliot,
>>>> I use OSX primarily (10.14.6 Mojave) and do not observe any CMD+key problem.
>>>> Anyway, image should not use charCode, because:
>>>>
>>>>     ActiveHand keyboardInterpreter
>>>>
>>>> should be:
>>>>
>>>>     a MacUnicodeInputInterpreter
>>>>
>>>> and MacUnicodeInputInterpreter does use the utf32 code
>>>>
>>>>     keyValue := evtBuf at: 6.
>>>>
>>>> Even if it would use evtBuf at: 3, that would only become a problem for accents é è or other non ASCII characters.
>>>> I have put Multilingual-nice.252 in inbox if ever we choose the modified VM.
>>>>
>>>> Did you build the VM from clean working copy or downloaded from Travis, or modified working copy?
>>>> Could you share what the image side duplicate/swap control/alt preferences are? ActiveHand keyboardInterpreter?
>>>>
>>>>
>>>> Le ven. 27 déc. 2019 à 06:35, Eliot Miranda <[hidden email]> a écrit :
>>>>>
>>>>>
>>>>> Hi Nicolas,
>>>>>
>>>>> On Thu, Dec 26, 2019 at 8:30 PM Eliot Miranda <[hidden email]> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Thu, Dec 26, 2019 at 12:21 AM Nicolas Cellier <[hidden email]> wrote:
>>>>>>>
>>>>>>>
>>>>>>> Let me suggest f70cf5f34041e0e197542651e90505a0b86325f6
>>>>>>> Please test it! (once the CI has built...)
>>>>>>
>>>>>>
>>>>>> My testing shows a serious regression with Apple/Command keys on MacOS.  This commit:
>>>>>>
>>>>>> commit a54a2240565e729b5f613a1c2b3277bb3a8a2873
>>>>>> Author: Nicolas Cellier <[hidden email]>
>>>>>> Date:   Mon Dec 23 21:05:54 2019 +0100
>>>>>>
>>>>>>     Convert charCode to latin1 encoding rather than macRoman on iOS.
>>>>>>
>>>>>>     Why? because image side uses unicode encoding for characters, and that matches latin1 up to 256.
>>>>>>     Once upon a time, the image side did use macRoman, so that explains the legacy code.
>>>>>>     But such legacy has become a drag, let's get rid of it!
>>>>>>
>>>>>>     TODO: same changes required on legacy Mac OS
>>>>>>     But I might apply those change after merging the compile_legacy_Mac_OS code in order to avoid potential conflicts.
>>>>>>
>>>>>> affects this file: platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m
>>>>>>
>>>>>> So we need to use an earlier version of the keyboard support and do a different merge to pull in Levente's changes.  Or we need to fix this.
>>>>>
>>>>>
>>>>> Well, the regression occurs, but it is nothing to do with platforms/iOS//vm/OSX/sqSqueakOSXApplication+events.m.  Reverting that does not fix the problem.  So I don't know what has caused the regression.
>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Le jeu. 26 déc. 2019 à 08:48, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>
>>>>>>>> I also want to fix https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/456 at least for the letters (ctrl+a to ctrl+z)
>>>>>>>> I find current Cog VM quite not usable on linux (OK, I am testing via Parallels virtualization with a Mac keyboard, but still).
>>>>>>>> I can propose a fix this morning.
>>>>>>>>
>>>>>>>> Le mer. 25 déc. 2019 à 23:36, Nicolas Cellier <[hidden email]> a écrit :
>>>>>>>>>
>>>>>>>>> Hi Eliot,
>>>>>>>>> I suggest we merge https://github.com/OpenSmalltalk/opensmalltalk-vm/pull/458 as Levente asked for.
>>>>>>>>>
>>>>>>>>> Le mer. 25 déc. 2019 à 23:30, David T. Lewis <[hidden email]> a écrit :
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Wed, Dec 25, 2019 at 10:43:41PM +0100, Fabio Niephaus wrote:
>>>>>>>>>> >
>>>>>>>>>> > On Wed, 25 Dec 2019 at 10:32 pm, Eliot Miranda <[hidden email]>
>>>>>>>>>> > wrote:
>>>>>>>>>> >
>>>>>>>>>> > >
>>>>>>>>>> > > Hi All,
>>>>>>>>>> > >
>>>>>>>>>> > >     so it seems things are stabilizing. The CI builds.  The recent commits
>>>>>>>>>> > > from Nicolas re events seem like goodness to me.  Are we agreed that I
>>>>>>>>>> > > should tag the a4a08f98c17b23ad54b5e361ffc0280e2ea8e0c3 commit or does
>>>>>>>>>> > > anyone suggest an earlier one?  This commit has VM sources that are
>>>>>>>>>> > > unchanged since Pablo Tesone's fix for gcc compilation of immediate floats.
>>>>>>>>>> > > It includes the socket name lookup fixes Levente wanted, and it has very
>>>>>>>>>> > > localized but useful fixes for events.  My vote would be to choose this.
>>>>>>>>>> > >
>>>>>>>>>> >
>>>>>>>>>> > +1
>>>>>>>>>> >
>>>>>>>>>> > Fabio
>>>>>>>>>> >
>>>>>>>>>>
>>>>>>>>>> +1
>>>>>>>>>>
>>>>>>>>>> Dave
>>>>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> _,,,^..^,,,_
>>>>>> best, Eliot
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> _,,,^..^,,,_
>>>>> best, Eliot