Excellent! Stef
On Jan 1, 2012, at 10:36 AM, Krishsmalltalk wrote: > > >> What I would like is that you take each of these points and send one mail >> for each with a detailed description. so that we can answer, turn them into >> bug entries.. > > > I do intend doing this in a while thru the week ahead. > > > Thx,... For other answers... > > > On Jan 1, 2012, at 1:58 PM, Stéphane Ducasse <[email protected]> > wrote: > >> >>> Wishing all Pharoer's a very warm n happy new year >>> >>> Since last few weeks, I have been experimenting with Pharo as a tablet >>> interface inside a tiny core Linux. The dev though happens on ubuntu. Some >>> more playing around for a group of trainees with pharo on windows 7, >>> platform I do not very much like. >> Sounds really interesting. >> Tell us more. >> >>> >>> How far is it a nice to have framework that provides for a complete UI, IDE >>> like a tablet interface and behaves akin to android that has it's own >>> AppStore, and application launcher in full screen on icon click. >>> >>> Anyone can download the base, and proceed to download the app source, files >>> archive thru an app installer, that verifies , validates and installs as >>> well as creates a launcher icon ( I will provide the screenshots once I >>> have them fine tuned) >>> >>> Rest of ideas and process are similar to android marketplace or iStore, >>> with definite controls on absolute stability and experience of the apps >>> >>> Some observations i will like the veterans to clue me on forward >>> >>> . I find Pharo 1.2 more stable than 1.3 latest... >> >> - Can you give some specific examples? Because like that this this is >> useless. >> >>> . Is morphic capable of supporting error free runtime GUI >> >> Yes >> I do not see why it would not. Now if you are changing and adding method >> with funny behavior then it makes sense that you get >> strange red square. >> >>> I enjoy it's capabilities, but feel it demands more intrinsic Pharo, >>> debugging ability to keep hacking n fixing. Crashes , damaged rectangle >>> painting are just few of the troubles >> >> When I use Pharo and I just code my application without tweaking and >> cleaning Morph I do not get damaged rectangles. >> I imagine that Polymoprh/pinesoft, DRGeoII…. people would not deliver >> production software if this would be the case. >> >>> >>> I expect a simple answer, not necessarily a longer explanation of >>> morphic and it's issues with global coordinates, polling event handling >>> which are nuisances. I have read thru as much of links, docs on morphic as >>> i could find, little working on it, but not enough to get an expert hang of >>> it. >> >> For polling we can now switch to events. After two years the windows VM >> got the input semaphore fix integrated. >> We will probably do that once we find a problem we saw while rewriting >> completely the event generation in HandMorph. >> >> >>> . Is system browser the best to study for morphic or should I look through >>> any other package..Debug n learn.. >> NOOOOOOOOOO! >> >>> . Simple way to get an embedded submorph from it's chain by it's name, >>> #subMorphNamed: does help but only in a limited way.. For those that it >>> digs. >> >> Named morph and looking for them is a bad concept. >> Instance variables is the way to go. >> >> >>> . Simpler way to bar extensions from a contributed older package >> every package can add its own menu entry. >> >>> , overriding base class methods, not really the final as in java but at >>> least prevent overwriting methods. A listing post file in of uncompiled >>> methods of this kind. Hacking in FilePackage does help.. But anything >>> better and more complete...and part of a base system. >> >> Sorry I could not get what you mean. >> >> >>> . Mechanism to autosave a defined package every method commit to a >>> packagename.mcz by default to ease crash recovery, use monticello only when >>> I need to version it as a done base. >> This is done. You have a crash recovery >> Any change you do is logged. >> >> >>> . Event handling of mouseLeave mouseEnter of a PluggableTextMorph, the >>> on:send: did not work, not that I have dug into it... But would like to >>> link it to the models methods directly in the builder method interface like >>> we do for PluggableButtonMorph click…. >> >> You should provide a real example because we do not have a crystal ball. >> >> >>> . Color of a morph is not updated when selecting it thru the halo. Minor >>> hack i have here, to get thru. >>> >>> . Why is extent attribute not updated when we use addMorph:fullFrame: >>> which changes the bounds for the submorphs in some cases at least as i have >>> noticed in debug step thru. >>> >>> . What are clear do not do rules in using morphic framework as it exists >>> today, as people have already experienced and attempted. >> >> I would love to know that too :) >> First basic OO programming. >> >> >>> . Debugger stepping through has most times an out of step highlighting. Any >>> fix on this.. >> >> This is a pending bug and when Athens is out, Igor will really look at >> this one. >> >> >>> . Settings window seems to have glitches in network settings, font updates >>> from OS... >> Be more precise. >> >> >>> . Autocompletion in windows seems to work on cr, but in Linux you have to >>> type thru. >> Be more precise >> >> >> >>> . The easiest way to have a core Pharo runtime image stripped of all dev >>> tools, classes, size is not the limitation here, but stable n bug free base >>> is a must. >> >> But there is no magic. Just work. And everybody can help. >> >> >>> . Tools n tips to make Pharo base with its loaded apps crash proof, just >>> isolate and close the app that causes the problem rest of the image should >>> continue to run. >> Yes that we want but this is really not simple. >> >> >>> issue not directly related: >>> . Windows support for OSProcess, i have read the threads and the note from >>> Eliot that it works on windows for cadence, but I cannot get it to work, >>> the ProcessWrapper works but is limited. If I do get it working does it >>> support interactive shell, viz i invoke a command that enters into it's own >>> loop of inputs n output till you send a quit. >>> >>> Lots more i will come up with in the next couple of weeks as I bring out >>> what i have. >> >> What I would like is that you take each of these points and send one mail >> for each with a detailed description. so that we can answer, turn them into >> bug entries... >> >>> >>> >>> Sudhakar krishnamachari >>> >>> Extn 91-40403012 >>> Cell 9902104814 >>> >>> >>> >> >> >
