On 9 August 2011 13:43, Nicolas Cellier <[email protected]> wrote: > 2011/8/8 Stéphane Ducasse <[email protected]>: >>> >>> >>> OK, some clean up was required, but you see, you can make things worse >>> when cleaning... >> >> Yes but in average this is really not the case. >> >>> It's like Pharo cleaning removed the antidote, but forgot to remove >>> the poison... >> >> Ok that kinds of things happens in ugly, undocumented, obsolete systems. >> >>> Now, of course, instead of a local polishing, it would be possible to >>> rewrite all. >> >> I'm starting to write tests for the events. Now since you are at it, I guess >> that you read this wonderful code >> full of hypothesis and other subtle strange behavior. So I'm quite sure that >> they will be problems. >> But we will fix them. Else we should stop now and do Javascript because at >> least we will look cool. >> > > Oh, If your goal is to remain popular, then indeeed, you should keep > away from changing such parts of the system which are not shining ;). > The obvious reason is that multilayer hacks with obfuscated contracts > are not maintanable... > so it's inevitable to create bugs in the transition and harvest complaints... > > Of course, if your goal is to make the system more simple and > maintanable, I'm with you (and I know it is). > But don't forget to extract the positive part of complaints that will > help you making the system better. > > Personnally, the finite state machine logic in Squeak/MouseOverHandler > is well beyond my understanding. > So I could not clean it easily myself, > > A cosmetic cleaning was attempted in Pharo.I have nothing against that. > I just pointed a problem in current cleaning, so it just have to be finished. > The stincky ivar mutation already was in Squeak. > I just said that it should raise an alarm in coder's eyes in the > future and I hope this helps. > Otherwise I can also shut up. > > The second approach suggested by Igor is more involving, consist in a > full rewrite, probably with these stages: > - gather description of the features to be supported, > - an inventory of third party interested in this kind of features, > - and from there, decline a new definition of contracts. > >>> But please, don't do it blindly and loose 2 third of features like the >>> new implementors window :( >> >> We are not doing things blindly: >> >> First it costs us money, time and we should all the lucky that Ben >> was really good and enjoy rewriting us boring UI >> without a decent widgets set (because he fixed them also the >> multiselection and others). >> Benjamin I publicly thank you for all the good stuff you did!!!! >> >> >> Second, what are these 2 thirds that we lost? Because two third is a >> lot. I'm dreaming to throw away two third of the stringHolder hierarCrap >> Please hightlight me. >> >> Sorry but I do not like this kind of statements because there are false and >> because we work hard to bring to life >> a system full of shit. Shit that was accumulated and of course stable >> because it did not move since years. >> In addition if everytime a guy spend 4 months to build a new tools we all >> complain (without constructive points to help him fixing that) >> better say to students that they are idiots and that we do not need their >> help. But in that case be ready to be with the same ugly system >> in the future. And we do not accept that. So there will be some glitches but >> we are stronger than them and we will fix them. >> >> Nicolas I'm quite sure that you would not say that around a beer to benjamin >> because I know you so pay attention because >> Ben is also one of the coolest students we got recently and we really liked >> his spirit and atttitude. Jumping in StringHolder and broken/duplicated >> widgets >> is not easy, nor it is to work on bootstrapping Pharo. >> >> Stef >> >> > > Sure, I don't want to restrain goodwill. > But as a user of the service I don't see all the sweat the hardworkers > put into the system, > I just see a regression that makes my life harder (I simply cannot > track the messages with the new windows). > And obviously, comparing the length of old menu with new one, two > third does not seem exagerated to me. > But maybe you don't work with those tools and install OB/RB after each > update... > > Nicolas >
Nicolas, it would help if you mention what exactly you missing instead of 2/3 :) -- Best regards, Igor Stasenko AKA sig.
