Agree with everything you said. As i said, my impression was that Guille pushed unfinished/incomplete pieces. And that is what i against. Because if we go that way, then someone will start pushing changes which even don't compile.
I hope i made my point clear, what blessed means. It is not synonym to 'stable'. On 7 August 2013 17:11, Esteban Lorenzano <[email protected]> wrote: > IMO is no point on ask for keeping an "all-time-stable" VM in the development > process (it is even a contradiction). > Is the same as asking for a 3.0 stable before release (yes, is the same... we > are trying to have same process for both). > > VM development is not stable right now, same way it is not stable 3.0 > Want stable stuff? you need to take the stable VM and Pharo 2.0 (even with > the bugs we all know). > Ok, we (the pharo developers community) cannot do that, we need to work on > the bleeding edge and because of that, some times we get uncomfortable > situations. > But then we all know how hard is to have a functionality working properly and > well tested. Our only way to do that is to put it in the development trunk > and ask for feedback. > > So... yes, I'm sorry it annoys some time to time, but latest-VM will have > periods in which it doesn't work as expected... is the only way this small > community can advance: Developers introduce a functionality, community tests > and provide feedback, then loop. > > Now, what we should have is a better way of mark stuff.. so we can backport > stuff et all. > For that, what we should have is: > - all sources in just one repository, including vmmaker (I'm working on that) > - create a tag (that we can branch and backport stuff) each release. > > no need to stress about :) > > my 2c > > Esteban > > > On Aug 7, 2013, at 4:43 PM, Igor Stasenko <[email protected]> wrote: > >> On 7 August 2013 15:55, Guillermo Polito <[email protected]> wrote: >>> >>> On Wed, Aug 7, 2013 at 2:41 PM, Igor Stasenko <[email protected]> wrote: >>>> >>>> My understanding how it should be: >>>> - you work in own branch (and in git, repository clone is already a >>>> branch, just on your computer). >>>> If you want to make it public you need to make own branch on gitorious >>>> and commit there. >>>> - once you finished with changes and tested it , you are free to merge >>>> with blessed >>>> (no need to have GODs permission to do that, or make pull requests) >>> >>> >>> We did all that. We tested it with Ben because he needed the change to send >>> ctrl events from the vm to the image, which were not sent. We tested in our >>> computers, and we had no much trouble. >>> >>> - I do not use delete key because I don't have one >>> - circumflex accent (return) was working with my keyboard layout >>> >>> So the vm was to me in a completely usable state. >>> >>>> >>>> >>>> now i understand we are humans and not immune to mistakes, but i >>>> insist that committing to blessed >>> >>> >>> 'blessed' is a crappy name for a repository. And the idea that the code in >>> this repository is always working flawlessly is nonsense. For example, we >>> had for months the problem with the smallinteger in there. >>> >>> And then, the unstable is in another repository then? Where is that >>> repository? Why do we have to make another repository for that and not just >>> a branch (in the same repo, not in my machine)? >>> >> >> so what you proposing? lets everyone put own unfinished changes into >> same repository >> without synchronization and see how it flies? >> it won't fly, i can tell you even without trying. >> >>>> >>>> should be done after some manual testing and checking it don't breaks >>>> things badly, and of course >>>> blessed is not for committing the unfinished code , which in the >>>> middle of work. >>> >>> >>> First, we considered it as finished. What you found was a bug, not a pending >>> to do task. >>> >> >> Ah, in that case it is perfectly ok then. >> It is your messages in previous mails gave me impression that it is >> 'not done yet' >> but 'work in progress' and change is not finished. >> >>> Second, I have not the time to change to and learn every keyboard layout to >>> test, nor to plug every external keyboard and see how it behaves. So if you >>> think we should have done that, it will never be finished, and we will die >>> with what we have. >>> >>> Third, we need an unstable official branch for the pharo vm code for people >>> to test. To me, as blessed is the only repository out there, it is the only >>> place where I can put it. >>> >> >> It is blessed in terms that every change you pushing there is blessed >> by you (as developer), >> which means "i recommend to use new version by others". >> And until you can bless your own changes you should not push them into >> this repository, >> while obviously you free to push/commit them elsewhere. >> >>>> >>>> >>>> That is my only issue , that you committed code which is not yet >>>> finished, and makes VM unusable, >>> >>> >>> Now again. To me the latest vm was completely usable. How can I prove that >>> the change is then finished if I do not test with every existing device and >>> layout? >>> >>> So people has to test it. And let's be realistic: given the few people using >>> nowadays the latest pharo vm, if I put my code in my own personal >>> repository, how many people will download my experimental VM to test it from >>> my hidden branch in my hidden and personal repo? I guess almost none. >>> >>> No one tests => never proved => never finished => never merged => dies >>> >> Agreed. >> >>> That does not work. >>> >>> So if bleesed is the stable branch, which is the latest unstable branch? Do >>> we need many repositories for that? >>> >> >> it is not stable branch, it is development. >> stable versions come time to time, we can tag them as stable. >> >>>> and we cannot release VM with important fix to smallinteger bug before >>>> you finish, which turns yourself >>>> into a bottleneck. >>> >>> >>> You can always go to the vm log >>> >>> https://gitorious.org/cogvm/blessed/commits/master >>> >>> and check my last commits and try to rollback them in any case. This change >>> was only two lines of code. I'm of course not the code owner. >>> >>> >>> Now, I'm tired to argue. I want to solve this by stating: >>> - how do we get the latest unstable and where should the code be >>> - how do we get the latest stable and where should the code be >>> >>> Something else is that I have the feeling that we are discussing because of >>> the stupid "blessed" name. So I did this: >>> - create a new pharovm repo inside https://gitorious.org/cogvm, the old >>> blessed repo stays the same >>> - I copied everything from the repo blessed to pharovm, including history >>> - I created a blessed branch, which should in the future have the properties >>> you like <=> latest stable >>> - let master for latest bleedingEdge (not experimental, but maybe buggy) >>> >>> And then if we want other branches we create them. >>> >>> https://gitorious.org/cogvm/pharovm >> >> just one question: why adding 'blessed' branch if you think it is stupid? >> name it stable then, since you don't understand/value of blessing or >> consider it is stupid >> and unnecessary. >> but again, we don't need 'stable' as separate branch.. makes little sense, >> since we can just tag certain versions as stable, like everyone else does. >> >> >> -- >> Best regards, >> Igor Stasenko. >> > > -- Best regards, Igor Stasenko.
