Smalltalk › Squeak › Squeak VM

Re: [Pharo-project] The roadmap aka fight plan to make image control host window

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

5 messages Options

Eliot Miranda-2

Re: [Pharo-project] The roadmap aka fight plan to make image control host window

On Sat, Sep 22, 2012 at 11:22 AM, Igor Stasenko <[hidden email]> wrote:

First, we need to make one thing:

Change the VM to not call ioProcessEvents function directly, but
rather use a variable which will hold a function pointer.
Initially, when VM starts, that function pointer is NULL, and so,
"process events" is NOP.

Please don't reinvent the wheel. The Cog source has a variable inIOProcessEvents that controls this. If negative it disables inIOProcessEvents. If non-negative it prevents reentrancy, essential if one is receiving events from the OS. There's a primitive, primitiveEventProcessingControl, to flip it between the enabled and disabled states.

This code has been used for years in the Newspeak system which has a full native GUI on Windows.

Going your own way creates an unnecessary fork.

Then we need to implement a module which will manage host windows,
and once image tells it to create a window,
it can change the function pointer to ioProcessEvents to own function.

This change, would allow us to temporary disable any external event
processing (like in case of callbacks).

Second, we should move all event-related stuff to new plugin (for
instance, all definitions in sq.h
like sqInputEvent , ioSetInputSemaphore() ioGetNextEvent() etc).

For all UI-related stuff the image should communicate with a new
plugin, while VM should get rid of all knowledge of any existence of
UI events,
like input semaphores, primitives for fetching them etc.

Third, of course, all screen update/refresh primitives should be
deprecated, and image should use new primitives to update the contents
of host window(s).

--
Best regards,
Igor Stasenko.

--
best,

Eliot

Igor Stasenko

Re: [Pharo-project] The roadmap aka fight plan to make image control host window

On 25 September 2012 19:35, Eliot Miranda <[hidden email]> wrote:

>
> On Sat, Sep 22, 2012 at 11:22 AM, Igor Stasenko <[hidden email]> wrote:
>>
>> First, we need to make one thing:
>>
>> Change the VM to not call ioProcessEvents function directly, but
>> rather use a variable which will hold a function pointer.
>> Initially, when VM starts, that function pointer is NULL, and so,
>> "process events" is NOP.
>
>
> Please don't reinvent the wheel. The Cog source has a variable inIOProcessEvents that controls this. If negative it disables inIOProcessEvents. If non-negative it prevents reentrancy, essential if one is receiving events from the OS. There's a primitive, primitiveEventProcessingControl, to flip it between the enabled and disabled states.
>

Yes i seen that.
But apparently, i don't want to just disable it by putting NULL, but
also i want to replace it (or wrap it) with own io handler (see
NativeBoost ;). For that, i need the function pointer.

> This code has been used for years in the Newspeak system which has a full native GUI on Windows.
>
Does that means that implementation is completely on language side? If
so, then why you need the rotten window management code in VM?
Lately, i seen Mars running native GUI (browser window) using Cocoa on
macs and GTK on linux.
(but you still have VM window hanging around, and you cannot get rid of it).

> Going your own way creates an unnecessary fork.
>
I would love to not reinvent it, but the user input handling is one of
the weakest spots in current VM implementaion.
I don't like many things there:
- VM is in control whether to create window or not
- VM in control what attributes window has
- VM in control how to translate OS (native) events to
"squeak-compatible" events, often losing important bits and making
everything crawl like hell
- VM in control when to update/refresh a window
- VM decides whether it run headless or not
- VM decides when to handle input events
list can be continued..

Personally, i don't like that image assuming that VM has to provide a
single "default" window
for it. It is image, which should manage that (if it wants to), but not VM.

--
Best regards,
Igor Stasenko.

Eliot Miranda-2

Re: [Pharo-project] The roadmap aka fight plan to make image control host window

On Tue, Sep 25, 2012 at 3:13 PM, Igor Stasenko <[hidden email]> wrote:

On 25 September 2012 19:35, Eliot Miranda <[hidden email]> wrote:
>
> On Sat, Sep 22, 2012 at 11:22 AM, Igor Stasenko <[hidden email]> wrote:
>>
>> First, we need to make one thing:
>>
>> Change the VM to not call ioProcessEvents function directly, but
>> rather use a variable which will hold a function pointer.
>> Initially, when VM starts, that function pointer is NULL, and so,
>> "process events" is NOP.
>
>
> Please don't reinvent the wheel. The Cog source has a variable inIOProcessEvents that controls this. If negative it disables inIOProcessEvents. If non-negative it prevents reentrancy, essential if one is receiving events from the OS. There's a primitive, primitiveEventProcessingControl, to flip it between the enabled and disabled states.
>

Yes i seen that.
But apparently, i don't want to just disable it by putting NULL, but
also i want to replace it (or wrap it) with own io handler (see
NativeBoost ;). For that, i need the function pointer.

Unlikely. Better to call the relevant function through the FFI. If you're implementing a native gui you don't need the VM involved in deciding when to poll for events.

> This code has been used for years in the Newspeak system which has a full native GUI on Windows.
>

Does that means that implementation is completely on language side? If
so, then why you need the rotten window management code in VM?

One doesn't need it. It's there and lies unused.

Lately, i seen Mars running native GUI (browser window) using Cocoa on
macs and GTK on linux.
(but you still have VM window hanging around, and you cannot get rid of it).

> Going your own way creates an unnecessary fork.
>

I would love to not reinvent it, but the user input handling is one of
the weakest spots in current VM implementaion.
I don't like many things there:
- VM is in control whether to create window or not
- VM in control what attributes window has
- VM in control how to translate OS (native) events to
"squeak-compatible" events, often losing important bits and making
everything crawl like hell
- VM in control when to update/refresh a window
- VM decides whether it run headless or not
- VM decides when to handle input events
list can be continued..

Personally, i don't like that image assuming that VM has to provide a
single "default" window
for it. It is image, which should manage that (if it wants to), but not VM.

You don't. But you don't need two VMs either. The existing VM windowing code can coexist with the native GUI code. In fact one can use the old code to debug the new code, and if you're Vassili you can switch an open window from Squeak window to native window and back :)

--
best,

Eliot

Igor Stasenko

Re: [Pharo-project] The roadmap aka fight plan to make image control host window

On 26 September 2012 00:47, Eliot Miranda <[hidden email]> wrote:

>
>
>
> On Tue, Sep 25, 2012 at 3:13 PM, Igor Stasenko <[hidden email]> wrote:
>>
>>
>> On 25 September 2012 19:35, Eliot Miranda <[hidden email]> wrote:
>> >
>> > On Sat, Sep 22, 2012 at 11:22 AM, Igor Stasenko <[hidden email]> wrote:
>> >>
>> >> First, we need to make one thing:
>> >>
>> >> Change the VM to not call ioProcessEvents function directly, but
>> >> rather use a variable which will hold a function pointer.
>> >> Initially, when VM starts, that function pointer is NULL, and so,
>> >> "process events" is NOP.
>> >
>> >
>> > Please don't reinvent the wheel. The Cog source has a variable inIOProcessEvents that controls this. If negative it disables inIOProcessEvents. If non-negative it prevents reentrancy, essential if one is receiving events from the OS. There's a primitive, primitiveEventProcessingControl, to flip it between the enabled and disabled states.
>> >
>> Yes i seen that.
>> But apparently, i don't want to just disable it by putting NULL, but
>> also i want to replace it (or wrap it) with own io handler (see
>> NativeBoost ;). For that, i need the function pointer.
>
>
> Unlikely. Better to call the relevant function through the FFI. If you're implementing a native gui you don't need the VM involved in deciding when to poll for events.
>

The idea about function pointer was to move ioProcessEvents into
separate plugin. Like that, if you don't build VM with this plugin,
you automatically have no UI and no ioProcessEvents, because there's
nobody who registers that function.
And then platform VM code is no longer obliged to implement that function.
For me it is first logical step in decoupling GUI handling from the
core VM feature set.
While having the flag simply disabling that function from being called.

>> > This code has been used for years in the Newspeak system which has a full native GUI on Windows.
>> >
>
>
>>
>> Does that means that implementation is completely on language side? If
>> so, then why you need the rotten window management code in VM?
>
>
> One doesn't need it. It's there and lies unused.
>

So why you need to keep something which is unused?
Especially when it is flawed, limiting and plainly wrong?

>>
>> Lately, i seen Mars running native GUI (browser window) using Cocoa on
>> macs and GTK on linux.
>> (but you still have VM window hanging around, and you cannot get rid of it).
>>
>> > Going your own way creates an unnecessary fork.
>> >
>> I would love to not reinvent it, but the user input handling is one of
>> the weakest spots in current VM implementaion.
>> I don't like many things there:
>> - VM is in control whether to create window or not
>> - VM in control what attributes window has
>> - VM in control how to translate OS (native) events to
>> "squeak-compatible" events, often losing important bits and making
>> everything crawl like hell
>> - VM in control when to update/refresh a window
>> - VM decides whether it run headless or not
>> - VM decides when to handle input events
>> list can be continued..
>>
>> Personally, i don't like that image assuming that VM has to provide a
>> single "default" window
>> for it. It is image, which should manage that (if it wants to), but not VM.
>
>
> You don't. But you don't need two VMs either. The existing VM windowing code can coexist with the native GUI code. In fact one can use the old code to debug the new code, and if you're Vassili you can switch an open window from Squeak window to native window and back :)

And, what will prevent you from using new API for GUI handling,
provided by VM plugin do to the same what Vassili did?
And is there a point in future, where you can say "okay, let's throw
this out, because it rotten to death and obsolete and there's far
better alternative(s)?" Or there's no such point exists?

You must know well, that at certain point, if you need to reach higher
quality levels in doing something, you must rethink, what tools you
will be using (or create new, finer tools). One cannot do DNA
engineering using magnifying glass, one needs microscope.

Unless we want to prevent people from even considering using smalltalk
for something more serious than weekend hobby experiments.
Because keeping low-quality infrastructure will make sure it will stay
like that.

Well, maybe you right, maybe this stuff is not worth touching, but
instead it is better to think how to completely circumvent it and let
it rot there.
But personally, i won't feel happy knowing that VM contains code which
nobody uses and sits there waiting for someone who will attempt to run
20-years old images.
For me it doesn't makes sense: if you want to run 20-years old image,
use 20-years old VM.
Compatibility problem solved.

--
Best regards,
Igor Stasenko.

Igor Stasenko

Re: [Pharo-project] The roadmap aka fight plan to make image control host window

Remember, Andreas added a nice new feature into (at least) windows
VMs, where upon startup it showing a splash image?
By reading a .bmp file and delaying rest of process by couple seconds.
That maybe useful. But what if tomorrow, my customer will want to keep
showing it, till user click on it?
What you will do? Right! Go and hack VM. And what if he will want to
show splash and progress bar while loading/initializing stuff? Another
shitloads of code into VM?
And something fantastic, what if customer would like to use different
image format for splash screen, png, jpeg? And what if we would like
to automatically download the splash image from web site before
showing it?

So, where is the end of this? When you will finally say "enough", we
need proper window management in image?

--
Best regards,
Igor Stasenko.