[Mm10s] 2019-12-16

### Last week:

My report, some are the same than Esteban, as we work a lot together :D

### Last week:

- (Pair Programming with Esteban) Modifying the TFFI callbacks to
include information about the Smalltalk block in the plugin data
structure. This improves the debugging experience and the detection of
errors when a Callback in the image side is garbage collected.

- (Pair Programming with Esteban) Compiling the Queue based
ThreaddedFFI plugin to use the latest libffi version. This is required
to build in Catalina.

- (Pair Programming with Esteban) Adding synchronization code to
correctly interrupt the AIO waiting idle loop. This is still work in
progress as we have to test more scenarios. This modification allows
us to extend enormously the idle time of the relinquish processor
primitive without losing events nor delays.

- (Pair Programming with Esteban) Fixing a glitch in GTK with the
keyboard input and event handling when the GTK windows opened by Spec2
is the topmost of the application

- Detection and fix of the GCC 8.3+ optimization glitch that prevented
the built of the VM in newer Linux distributions.

- Another iteration on the FreeTypeFont glitch issue.

- Detecting all the requirements to use the Sista bytecode in the
default image. This modification requires some changes in Fuel. There
is a need for more changes in the debugger if we want to integrate
full block closures.

- Fixing small issues and reviewing PR for Pharo8

- Supporting Students in their modifications of the VM and experiments
they are doing (Theo & Pierre here in Lille, and Federico in Buenos
Aires): if you have problems just call :D.

### This week (starting 2019-12-16):

- Pushing Pharo 8 release (Release Notes / PR / fixes)

- Adding tests to the interruptable AIO and main thread worker of the
headless VM.

- Clean-up of code of the VM to allow to compile with MUSL and
different libc implementations. This will allow us to run in Alpine
Linux, improving the experience of Pharo running in containers. For
doing so the correct way is to extend the debugging functions to allow
different streams and outputs (today it is done through a hack, so we
have to keep the behavior but implemented in a way that does not break
the encapsulation of libc).

- Working on GC tests with Guille.

On Mon, Dec 16, 2019 at 11:00 AM Esteban Lorenzano <[hidden email]> wrote:

--
Pablo Tesone.
[hidden email]