On 08/06/2011 04:00 AM, Tomasz Finc wrote:
* the promised things like regular ZIM dumps sohuld be
finally done by
WMF to support the spread of the software / project
: The current XML dumps is system is fairly broken and has been a
nightmare to manage over the last couple of years. And if you think
the XML system is bad then don't even look at the HTML generation
system. Instead of pushing more work into it we've opted to go for a
much simpler route. Allow the collections infrastructure to generate
full project files. Using the existing system we make the workflow
much easier for anyone wanting openZIm files while at the same time
not making our infrastructure more complicated then it needs to be.
I agree, this is maybe the simplest way today... but the time we may
save now will have to be invest/waste in the future.
So, after many discussions, also in Wikimania, I'm really not convinced
by this approach and I think this is not the best way.
The way which is really efficient and sustainable is to work on DumpHTML:
* This is the faster solution
* This is trivial to deploy for Mediawiki admins (should also be the
case at the WMF)
* This is not too expensive to develop
* This allows to make sophisticated post-treatment if necessary
* You may modify the rendering in details (we work directly with the
mediawiki render engine)
* This is a simple and maintainable (works with hooks and skins) solution
That's the reason why I will propose a grant request to a chapter to get
that done for the summer 2012. Preparation of the grant request is
already started, feel free to make comment on the talk page:
http://www.kiwix.org/index.php/Mediawiki_DumpHTML_extension_improvement#Wor…
* openZIM and Kiwix are hardly able to cope with the
expectations due
to missing developers - not only software-wise but also concerning
documentation
Totally agree. We've made some amazing progress on both the Kiwix
betas and Collections extension but we can still do more. I think we
have to look at the engagement that we do (LinuxTag, Wikimania, Berlin
Dev Conf, beta testing, blog posts, etc) and really assess whats
working and whats not. The chapters can do a lot here to fund
initiatives and help run them.
I think, what is urgent is to find a solution with the WMF for openZIM
dev. ressources.
For Kiwix, I think what is essential is to get the chapters involved.
So, if the WMF continues to support Kiwix dev. in the manner they do
today, this is perfect.
* reimbursement for the server to build ZIM files
+1
Ok, what would be the best way to request that. At Kiwix we have two
servers;
* one rented for the online presence
* one in my guestroom to build the ZIM files. This is a server with a
disk array.
We speak her of the second one (I already get money for the first one).
It costs around 5000 euros.
* house the Kiwix server in a NOC
+1 .. I've said for a while that we should just move it over to
download.wikipedia.org
What is the NOC? I think we have to keep it in Europe and at the best in
the neighborhood of Switzerland.
* writing proposal for a contractor on openZIM
Looking
overhttp://openzim.org/Roadmap its bit hard to know what's
happening post Wikimania. Whats the future look like?
I could prepare something if needed?
Regards
Emmanuel