Hi all,
Wikimedia is at 1.20/wmf10 now. That means that it has been working with 1.20 alpha for the past 20 weeks. Isn't is about time we start preparing something usable called 1.20, 2.0, or whatever, for the outside world, too? Previous experiences tell us that getting something release ready takes at least 6 weeks, so we we'd want to have a stable release by end of October, we'll have to starting doing something very soon.
Cheers!
On 08/31/2012 03:19 AM, Siebrand Mazeland (WMF) wrote:
Wikimedia is at 1.20/wmf10 now. That means that it has been working with 1.20 alpha for the past 20 weeks. Isn't is about time we start preparing something usable called 1.20, 2.0, or whatever, for the outside world, too?
I've started discussing the tarball issue here (https://www.mediawiki.org/wiki/Requests_for_comment/Tarball_maintenance) this week. If we assume that this starts the 6 week cycle, then we should have a release during the first week of October.
Mark
Honestly, I'd love to see Daniel's password hashing system merged before the next release, if that's at all possible.
*--* *Tyler Romeo* Stevens Institute of Technology, Class of 2015 Major in Computer Science www.whizkidztech.com | tylerromeo@gmail.com
On Fri, Aug 31, 2012 at 9:29 AM, Mark A. Hershberger mah@everybody.orgwrote:
On 08/31/2012 03:19 AM, Siebrand Mazeland (WMF) wrote:
Wikimedia is at 1.20/wmf10 now. That means that it has been working with 1.20 alpha for the past 20 weeks. Isn't is about time we start preparing something usable called 1.20, 2.0, or whatever, for the outside world, too?
I've started discussing the tarball issue here (https://www.mediawiki.org/wiki/Requests_for_comment/Tarball_maintenance) this week. If we assume that this starts the 6 week cycle, then we should have a release during the first week of October.
Mark
Human evil is not a problem. It is a mystery. It cannot be solved. -- When Atheism Becomes a Religion, Chris Hedges
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Fri, Aug 31, 2012 at 6:29 AM, Mark A. Hershberger mah@everybody.org wrote:
On 08/31/2012 03:19 AM, Siebrand Mazeland (WMF) wrote:
Wikimedia is at 1.20/wmf10 now. That means that it has been working with 1.20 alpha for the past 20 weeks. Isn't is about time we start preparing something usable called 1.20, 2.0, or whatever, for the outside world, too?
I've started discussing the tarball issue here (https://www.mediawiki.org/wiki/Requests_for_comment/Tarball_maintenance) this week. If we assume that this starts the 6 week cycle, then we should have a release during the first week of October.
I think this is fine. The steps for making the tarball are here (including link to make-release): http://www.mediawiki.org/wiki/Release_checklist
I don't think we should make this the "2.0" release.
Sam would be the one to publish the tarball, but anyone can generate an unofficial alpha tarball, and I'd encourage that.
I'd like to get something finished in October (I was hoping September, actually, but events seem to have overwhelmed that). Getting things done by then means not waiting for anything that doesn't make it in soon. We should probably branch from the same branch point as 1.20wmf12.
Rob
On 08/31/2012 05:02 PM, Rob Lanphier wrote:
Sam would be the one to publish the tarball, but anyone can generate an unofficial alpha tarball, and I'd encourage that.
We can already use the code to create a tarball, we already have nightlies (https://www.mediawiki.org/wiki/Nightlies), and we already have the GPL license to distribute the code, so how do you see this working?
How would unofficial alpha tarballs carry any weight with anyone if there isn't a direct path from alpha to release? It seems like that is just encouraging forking, something that I'm trying to avoid.
Is this a quality or security concern? If so, could you clarify the risks you see?
I like Sam and think he is very responsive to the community, but this sounds like the Foundation is treating the tarball as a /direct/ interest when its interest in the tarball is an indirect one as you've said before in http://hexm.de/la (gmane link):
I've lobbed the idea out there of MediaWiki-focused fundraising to WMF people outside of WMF Engineering, to somewhat lukewarm response in the past.
The Foundation already accepts code contributions and allows code re-use through its source repository and the GPL license, so remaining the tarball gatekeeper seems unnecessary.
Why not allow the community to make a tarball release if the community cannot give money to support the tarball?
Mark.
Hi Mark,
Comments inline
On Fri, Aug 31, 2012 at 7:01 PM, Mark A. Hershberger mah@everybody.org wrote:
On 08/31/2012 05:02 PM, Rob Lanphier wrote:
Sam would be the one to publish the tarball, but anyone can generate an unofficial alpha tarball, and I'd encourage that.
We can already use the code to create a tarball, we already have nightlies (https://www.mediawiki.org/wiki/Nightlies), and we already have the GPL license to distribute the code, so how do you see this working?
How would unofficial alpha tarballs carry any weight with anyone if there isn't a direct path from alpha to release? It seems like that is just encouraging forking, something that I'm trying to avoid.
It's not forking. It's helping with the pre-release practice. If you're running the same scripts, and going through the same process we'd eventually go through, then what we're doing is simply a rubber stamp (and can probably just automate it at some point). This isn't much different than the process that some Linux kernel devs follow sometimes (e.g. Andrew Morton) prior to an official Linux release from Linus.
Why not allow the community to make a tarball release if the community cannot give money to support the tarball?
Because "the community" isn't going to produce the tarball. Someone in the community (you?) will be doing it, and I want whoever we hand the keys to build some trust in with everyone else that they are going to release a quality product before anointing them. Or rather, before I can recommend we do it; it's not solely my decision.
The step of actually uploading the tarball to the right place on mediawiki.org is the most mechanical (and thus least interesting) part of the process anyway. Authorship of the release notes, testing on various platforms with different databases, identifying and fixing the blockers, and applying the last bits of polish are the parts that generally take the longest. And, as near as I know, none of that is done, so let's focus on those bits.
Rob
On 09/04/2012 03:45 AM, Rob Lanphier wrote:
On Fri, Aug 31, 2012 at 7:01 PM, Mark A. Hershberger mah@everybody.org wrote:
On 08/31/2012 05:02 PM, Rob Lanphier wrote:
Sam would be the one to publish the tarball, but anyone can generate an unofficial alpha tarball, and I'd encourage that.
How would unofficial alpha tarballs carry any weight with anyone if there isn't a direct path from alpha to release? It seems like that is just encouraging forking, something that I'm trying to avoid.
It's not forking. It's helping with the pre-release practice.
Agreed.
And, as you can imagine, when I have some time (this coming weekend?) I'll try to put together a alpha 1.20 tarball.
I want whoever we hand the keys to build some trust in with everyone else that they are going to release a quality product before anointing them. Or rather, before I can recommend we do it; it's not solely my decision.
Thanks for clarifying, this seems completely reasonable..
Mark.
wikitech-l@lists.wikimedia.org