I'm going to move all the syx platform to berlios.de. After all, I always
used it and never bothered me. In my opinion it's the more stable and
efficient free hosting out there.
Actually, we'll use berlios.de for the website and the downloads. The other
services will stay on their current providers as they're more user-friendly
than berlios.
This will happen very soon this week.
Any hints? What about the website? Any better hosting?
I'm sorry for the inconvenience.
Best regards,
--
http://syx.googlecode.com - Smalltalk YX
http://lethalman.blogspot.com - Thoughts about computer technologies
http://www.debian.org - The Universal Operating System
I can recommend Google's App Engine. Their free quota is big enough
for decently popular web sites, you can treat is a pure static html
site or add more advanced features with their sever-side python
processing. I've been using it for hosting http://forums.fofou.org/
for almost a year and my blog (http://blog.kowalczyk.info/) for
several months. It's free, it works, it's easy to update - I wrote
more about it at
http://blog.kowalczyk.info/article/App-Engine-as-generic-web-host.html
Krzysztof Kowalczyk
It says "Read about using appcfg.py to upload and deploy one." in the
dashboard but appcfgpy.html doesn't exist. This demonstrates again (after
SVN) that google is not such reliable. The only reason I'd like to port syx
to berlios is the 100% reliability of the service.
Another reason is that berlios offers cgi, so I was thinking it was possible
to run syx in there, but I'm not sure of.
Anyway I'll wait a couple of days.
The website URL will be syx.berlios.de.
And... yes, it's definitely messed up :)
The idea is to transfer everything to berlios
as it has mailing lists, git and the bug tracker. I'll port the current git
hosting to berlios as soon as I can, while the ML and the bug tracking can
stay there for both "compat" and user-friendness. In other words, syx is
currently known to be on googlecode, so for the next 2-3 releases we
can't completely drop google.
And think syx wouldn't have been such publicized and active (in the past
months) if it was first-released on berlios, so google gave a good sprint
for propaganda in my opinion.
[PS, some news from the object branch]
You know the object branch contains a new object structure, a new memory
management and therefore a new garbage collector. It's still under heavily
development but currently works good enough to be merged in the next release.
Things do get done are:
- Port the new memory management to windows
- Test processes serialization in the image, much critical
- Object finalization
- Improve get_class() performance
- Fix object/pointer recognizing
Objects are now allocated in a huge data stack instead of get malloc'd. The
garbage collector does mark and compact. Contributed code is welcome in
order to get the above things done for the next release.