2018-04-01 11:36 GMT+02:00 Stephane Ducasse <stepharo.self@gmail.com>:
Hi Torsten
Thanks for this question.
Ephemerons do not work in Pharo sadly and we know it. We cannot then take advantage of them. What you see is in fact a degradation of the situation around the VM. :( We are concerned about this but the path for our future in this domain is unclear.
Hi Stephane, the future is in your hand. Put milestones on opensmalltalk-vm, open new entries on opensmalltalk-vm issue tracker, put engineering force to solve pharo-specific build failures. In one word, help us to help you. So could you please explain what did you do exactly to help ephemerons being implemented in spur, and what are the forces couter-acting your efforts?
What is clear is that we will reduce the number of Plugin (as pablo did for freetype) and use FFI. But FFI callbacks are not really good. And we need a threaded VM. Now nothing happens on this front.
Yes, we all want that! It's just that it does not happen by magic nor handwaving.
Let us face it there is a bad energy around. After years of discussion our engineers are still struggling and slowed down because of the dispersion of plugin code. In addition, from what I understood there is no development branch :) amazing is not it in 2020!
How sarcasms are going to federate and encourage people to make progress? What can you obtain with this sort of sentence but further splitting the community? Is it your goal?
On another scale, since 1 year, I was focusing on making that our institute offers a permanent position to Clement but I failed because the forces against where too strong (politics) and clement should reapply next year. Now to improve his CV, clement is doing a postdoc at VUB and will start working on Truffle (the VM of Oracle) and continue to work part time on Sista. This is not that bad, like that he will gain experience on other systems and in the future we will be in the position to see if there are real alternatives for us. May be writing a new VM on top of other existing infrastructure is the way to go. We will see in the future.
Unfortunately, the rules of French institutions certainly do not help you... Maybe Truffle is a possible future for you, more solid than 100% own fork for sure. To be fully transparent
- we will have one phd working on IoT and how to reduce/clean the image and VM size and many other points she will probably work on a fork because I cannot suicide a student (by asking her to understand the messy plugin architecture) and she will need some space to experiment. Now if improvements she will perform take ages to be integrated in the main VM I will ask her not to do it and to focus on her future and science. I imagine that everybody paying attention to people can understand this.
Good to know that you are paying attention to people, I had the impression that this was not allways the case. I'm willing to help Pharo, I've helped producing Pharo brand 64 bits windows VM more than one year ago. I'm struggling to make the community beneficial to all, but I don't feel like I get a lot of support and wonder if I'm not completely mistaken when I see all the efforts deployed to tear it apart...
Stef
On Tue, Nov 7, 2017 at 9:00 PM, Torsten Bergmann <astares@gmx.de> wrote:
Hi,
by accident I stumbled upon the following code in AnnouncementSubscription:
makeWeak action isBlock ifTrue: [ self error: 'Not currently available due to missing ephemerons support']. ...
My understanding is that we have Ephemeron support in Pharo now working. Right?
Anyone with more insights who knows how to fix/complete this? Should we at least open an issue?
Thx T.