[ANN] Roadmap progress report November 2019

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

[ANN] Roadmap progress report November 2019

EstebanLM
Hello,

I am sorry for the delay, last week was very chaotic :)

Spec2
---
- Make SpComponentListPresenter to acquire "list properties":
        - selection (single, multiple)
        - it can respond to #contextMenu:
        - takeKeyboardFocus
- Lists/Trees/Tables now can be searched (still "standard" implementation, misses a search by model)
- Lists/Trees/Tables now can be sorted

Spec2 + Gtk
---
- [Worker VM] Running event loop on main thread while VM runs on worker thread.
- Modify startup to automatically decide if it can run GtkMainRunLoop or GtkPollingRunLoop, depending on where is running the VM.
- Make backend to properly defer executions to main thread when needed (implemented GtkRunLoop class>>defer: for that)
- Fix the trunk for windows (detected a problem while coexisting SDL2 and GTK3 that may cause the need to use exclusively one or the other).

TFFI
---
- Improve callout performance using a semaphore pool instead creating a new one each call

VM
---
- Work on Worker VM (allow VM to run in a worker thread).
        - Merged with headless trunk
        - Merged with AIO interrupt
        - Now can be tested (but still some interruptions missing)
- (Re)making an environment to build on linux (to test the problem while building Debian/Ubuntu modern distributions)
- Investigating the garbage collection corruption problem. Integrating proposed solution of Clément Bera.
- Fixing problems in main thread worker and interruptable async IO.

Others
---
- Voyage-UnQLite backend working with new TFFI (this is to make some testing around using several libraries/threads in a single image)
- ObjCBridge package working with new TFFI (for same reasons as before)

Smalltalks 2019
---
- Presentation in Smalltalks 2019 (Neuquén, Argentina): Contributing to Pharo, new things for Pharo 8.
- Meetings during Smalltalks.
- Presentation in universities and companies in Argentina.
Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Roadmap progress report November 2019

EstebanLM
Hi,

> On 9 Dec 2019, at 13:51, Alexandre Bergel <[hidden email]> wrote:
>
>> Hello,
>>
>> I am sorry for the delay, last week was very chaotic :)
>
> No problem! This is super cool!
>
> Your report is amazing! Keep going!
>
>>
>>
>> VM
>> ---
>> - Work on Worker VM (allow VM to run in a worker thread).
>> - Merged with headless trunk
>> - Merged with AIO interrupt
>> - Now can be tested (but still some interruptions missing)
>
> Just wondering. A worker VM will have its own image right? Can it be otherwise? I am just curious whether two VMs can run on the same image. Maybe this is a stupid question, but I am wondering :-)

Worker VM just describes the capability of the VM to run in a worker thread (which means we can keep the main thread to other stuff, like running the event loop).
Now, while I know there is some work/experiments to revive something like the hydra (running several images in one VM, or several copies of same image), I do not know the opposite :)

And this Worker VM (we call it “ogre”, because it has two heads :P) will work with the regular image, nothing changed there.

Esteban

>
> Cheers,
> Alexandre
>