Hi everybody, Aza, Atul, and I just had a lengthy conversation about the release plans for tomorrow, and we have decided to somewhat tweak the way we are doing this release. Ubiquity 0.5 will still be released tomorrow, but it will be released as a separate entity from 0.1.8, and will be called Ubiquity.Next or something along those lines. It will be on addons.mozilla.org under this separate name, so both 0.1.8 and 0.5 will be available there. The two versions of Ubiquity will be kept separate until 0.5 is in a state that we are ready to push on to all of our ~400,000 users. The requirements for this merge will be something like:
1) Have the top x percent of 3rd party commands converted to Parser 2 format 2) Fix the network call problem in a way that allows us to re-enable async network calls for noun first suggestions (through server-side caching or other means) 3) Resolve bugs by design that will surely be identified as we get a bunch of new users on Ubiquity.Next The reason we do not want to force everyone to upgrade to 0.5 at this point is because the benefits of 0.5 in it's current state don't necessarily outweigh the negative changes for many of our users. With the async suggestions basically disabled for noun first suggestions, the biggest positive changes made between 0.1.8 and 0.5 are the more sophisticated parser, resulting in the removal of hyphens, and internationalization. While this stuff is very cool, and moving in the right direction, 0.5 breaks all 3rd party commands and also breaks some user habits (like the way delimiters force you to type "weather near boston", whereas in 0.1.8 you can just type "weather boston"). Breaking user habits is ok if the positive gains outweigh the necessary changes in behavior. We think that 0.1.8 and 0.5 should be kept separate until it is clear that 0.5 is a valuable upgrade to nearly all of our users, which will be true when the requirements listed above are fulfilled. While this is a labs project, we also have hundreds of thousands of users who depend on Ubiquity to make their computing lives easier, so we need to be cautious about pushing major changes on everybody. Aza will be creating a screencast further describing the reason for maintaining the two separate extensions soon. How does this sound to everyone? Any questions or concerns? -Brandon On Sat, Jul 4, 2009 at 2:54 PM, Fernando Takai <[email protected]>wrote: > > I think Tuesday is a good day ;D > > Btw, i emailed the couchdb user-list, but i really need the logs from > the server (so they can say what's going on). > Also, i prototyped another bug-report app [1], using MongoDB [2] - if > we can't make couchdb work, i think this is a good replacement. > > [1] http://bitbucket.org/fernandotakai/ubiquity_bugreport/ > [2] http://www.mongodb.org > > On Jul 4, 2009, at 6:47 PM, Zach wrote: > > > > > What about folding the Ubiquity commands into the pre-existing add-ons > > site? Can we bring SUMO on board to get the documentation up to > > snuff? That would get us back on track much more quickly- and it's > > what is going to happen sometime down the line anyway. > > > > Wasn't there a plan to get someone working on the HERD full time for > > the summer- an intern or something... > > > > -Zak-the-guy-who-promises-to-do-more-work-instead-of-complaining-:-) > > > > > > > > On Jul 3, 8:38 pm, "mitcho (Michael 芳貴 Erlewine)" > > <[email protected]> > > wrote: > >> Jono, this sounds great. The server-side is definitely a priority > >> worth seriously working on in the near future. Looking forward to > >> seeing you on Monday! :D > >> > >> I will email ubiquity-i18n about the localization time schedule. I > >> personally think the more public outreach to the broader localization > >> community (through dev-l10n), especially given the limited time until > >> next Tuesday, is better suited to right after the 0.5 release. > >> > >> mitcho > >> > >> > >> > >> > >> > >>> Hey everybody, > >>> I apologize for my panicky email on Thursday afternoon; I was very > >>> frustrated at my inability to get the server-side components > >>> working, > >>> and I got pretty emotional (as those who have to share an office > >>> with > >>> me can attest). Anyway, after hearing from everyone about it, I > >>> think > >>> we are good to launch early Tuesday morning. > >> > >>> Today I pushed a 0.5pre4 (up on the site now). It includes > >>> changes to > >>> the web content and the Report Bug command, so that users who are > >>> looking to report bugs are directed to Get Satisfaction, and users > >>> looking for more commands are directed to the Commands In The Wild > >>> wiki page. I think both of these will be more helpful than > >>> directing > >>> them to non-functional Herd and Bug Reporter services. > >> > >>> I think 0.5pre4 is ready to be renamed 0.5.0 and released, but it > >>> should be stress-tested by everyone involved to make sure it can > >>> upgrade cleanly, that it has decent performance and generates > >>> reasonable suggestions, and that the tutorial works. > >> > >>> The Herd and the report-bug service will likely require a complete > >>> rewrite of the backend infrastructure before they are functional. > >>> They both depend on a single couchdb install which currently > >>> stands at > >>> 6 GB, and which cannot run for more than a few minutes without > >>> crashing. I was unable to get it to stay up for any length of > >>> time, > >>> even after spending most of yesterday on it. So I think we'll > >>> need to > >>> start some new server infrastructure from scratch in order to > >>> support > >>> Herd and the bug reporter the way we want to. That should be one of > >>> the main priorities after we release 0.5.0. > >> > >>> As I said, I think we're good to release Tuesday morning. I agree > >>> with Christian about pushing to get translations done for that. > >>> Please, no new features or functional changes: just translations, > >>> testing, and bug fixes. > >> > >>> Aza has agreed to take responsibility while I'm gone, to write the > >>> release blog post for 0.5 and to manage the release of any dot- > >>> patches > >>> that are afterwards found to be neccessary (0.5.1 etc). > >> > >>> I am leaving on my trip tomorrow morning, so I can do no more on > >>> this > >>> release. The rest of you will have to decide the exact timing of > >>> the > >>> 0.5.0 release, and manage its execution. > >> > >>> Good luck! > >>> --Jono > >> > >>> On Fri, Jul 3, 2009 at 9:19 AM, "mitcho (Michael 芳貴 > >>> Erlewine)"<[email protected]> wrote: > >> > >>>> As Jono already mentioned a number of the reasons I gave to > >>>> strongly > >>>> push for a release today, it suffices to say that I agree whole- > >>>> heartedly with Blair. Amen. > >> > >>>> I'm fine with a Tuesday release, but we should make it a firm > >>>> commitment and we should make it crystal clear on trac and/or the > >>>> list > >>>> (ideally both) *EXACTLY* what the blockers are that we are up > >>>> against > >>>> (if any). > >> > >>>> m > >> > >>>>> I think Personas has a dual release schedule - one for the client, > >>>>> one > >>>>> for the server-side stuff. It seems to work pretty well, and I > >>>>> don't > >>>>> see > >>>>> why that method couldn't be applied to Ubiquity. To me, client > >>>>> (Ubiquity) and server (Herd) are two separate products. > >> > >>>>> I don't think the release should be pushed back for new features. > >>>>> For > >>>>> serious bugs yes, but there's nothing major blocking that I know > >>>>> of. > >>>>> There's no such thing as a perfect release - there will always be > >>>>> something that could have been fixed, could have been better. > >> > >>>>> So I'm for releasing now. Or Tuesday if really necessary. But not > >>>>> any later. > >> > >>>>> - Blair > >> > >>>>> On 3/07/09 1:09 PM, Jono wrote: > >> > >>>>>> P.S. remember that if upgrading to 0.5 breaks a user's favorite > >>>>>> third-party command, they can always go back to parser 1 using > >>>>>> the > >>>>>> Settings page. > >>>>>> (Anybody doing support should keep this in mind as it will be a > >>>>>> common > >>>>>> complaint.) > >>>>>> --Jono > >> > >>>>>> On Thu, Jul 2, 2009 at 6:08 PM, Jono<[email protected]> > >>>>>> wrote: > >>>>>>> How about we release Tuesday? Mitcho points out the following > >>>>>>> reasons > >>>>>>> we should release after all: > >> > >>>>>>> 1. if we push it back we will be letting down all the people who > >>>>>>> have > >>>>>>> already blogged/written/read about the 0.5 release > >>>>>>> 2. the XPI itself is ready > >>>>>>> 3. The server components are not ready, but 0.5 without bug > >>>>>>> reporter / > >>>>>>> herd is at least not any worse than 0.1.8 without bug reporter/ > >>>>>>> herd > >>>>>>> 4. The rest of the community can handle support without me ;-) > >>>>>>> 5. Getting people onto 0.5 will reduce the load on the > >>>>>>> ubiquity.mozilla.com. > >> > >>>>>>> We should replace the herd and bug reporter pages with > >>>>>>> notifications > >>>>>>> that tell users those services are down for repair, and then > >>>>>>> focus > >>>>>>> on > >>>>>>> getting them working after the 0.5 release is out. > >> > >>>>>>> Atul and Aza have agreed to manage the release while I'm gone. > >>>>>>> --Jono > >> > >>>> -- > >>>> mitcho (Michael 芳貴 Erlewine) > >>>> [email protected] > >>>> http://mitcho.com/ > >>>> linguist, coder, teacher > >> > >> -- > >> mitcho (Michael 芳貴 Erlewine) > >> [email protected]http://mitcho.com/ > >> linguist, coder, teacher > > > > > -- > Fernando Takai > http://twitter.com/fernando.takai > > > > > > > > -- Brandon James Pung Massachusetts Institute of Technology Class of 2010 Department of Electrical Engineering and Computer Science 828.777.8640 | [email protected] --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "ubiquity-firefox" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/ubiquity-firefox?hl=en -~----------~----~----~----~------~----~------~--~---
