Hi Manu and Stefan, Don't worry about the open source test. We are committed to this. I will enter two 4.0 JIRAs shortly for the following and assign Continuent engineers to work on these tasks.
1.) Develop a simple MATs that can serve as the basis for minimal testing of check-ins. 2.) Implement CruiseControl for Sequoia (followed by Hedera, Bristlecone, and Myosotis). This will ensure that the code remains stable and egregious errors are quickly repaired. I would like to suggest we begin to move forward this week with Sequoia 4.0. Here are the steps from my point of view: 1.) Tag the current contents of CVS HEAD (actually I would like to create a branch for this) so that we can merge from it as needed later on. At this point 3.0 can be discarded. 2.) Merge 2.10 into CVS HEAD, fully replacing current contents of that codeline. 3.) Branch immediately to create a 4.0-STABLE branch. We will move our commercial development to this branch. 4.) Implement CruiseControl builds for CVS HEAD and 4.0-STABLE. 5.) Start to implement the GPL test suite. The main problem with the test suite is that it may take a little while to devise a good way to start and stop controllers and set up databases. We are developing a new management framework that may be useful for this. Please let me know how this plan suits you. Thanks, Robert On 4/2/08 4:45 PM, "Emmanuel Cecchet" <[EMAIL PROTECTED]> wrote: Hi Stefan, > First let me say Thank you Emmanuel for your great work in the last days > fixing some old bugs and applying some of the patches that were sleeping > in the Jira :-) I have hope that someday i can build Sequoia from the > SVN directly :-) > You are welcome. I think we have reached an agreement with Continuent so that we can have an open source version that moves forward with community contributions while still leaving the opportunity for Continuent to base products on stable branches and builds. The situation is still a little delicate with 2.10 since this is the product branch and commits have to be carefully reviewed first. > To the future plans. > I must stand up and raise my finger, i can't find any point about making > Sequoia stable. The last point in 4.x is the only thing that goes this > way. I hope i can help you as much as i can with the Test Suite and > writing Tests. > This was one of my concern too and that's why I did not want to start working on 4.x before being sure that Continuent would commit to contribute its test suite (at least partially) to the open source community. We will probably have to limit the number of tested features but the test suite will be key for this stability. > Another feature/bug* that would make Sequoia more stable is the > WAN-scenario that is not yet supported by Sequoia. > > Any Ideas? > Multi-master replication in a WAN scenario is unlikely to happen. Asynchronous replication between interconnected site is doable. I would recommend the reading of section 4.3.4.3 on Network Partitions from my last Sigmod paper (http://infoscience.epfl.ch/record/118488). The CAP principle (by Fox & Brewer) is a must read too. > * > https://forge.continuent.org/pipermail/sequoia/2007-September/005625.html > https://forge.continuent.org/jira/browse/SEQUOIA-980 > Ok, I have commented on the JIRA issue, you can have a look and let me know what you think. > https://forge.continuent.org/jira/browse/SEQUOIA-983 > I guess this one should be a quick fix. Need to see with Robert if he allows that in 2.10. > https://forge.continuent.org/jira/browse/SEQUOIA-984 > Commented the JIRA issue as well. IMHO, we should remove that feature... Thanks a lot for your feedback and support, Emmanuel _______________________________________________ Sequoia mailing list [email protected] https://forge.continuent.org/mailman/listinfo/sequoia -- Robert Hodges, CTO, Continuent, Inc. Email: [EMAIL PROTECTED] Mobile: +1-510-501-3728 Skype: hodgesrm
_______________________________________________ Sequoia mailing list [email protected] https://forge.continuent.org/mailman/listinfo/sequoia
