November 2009 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] accessors entry in submenu of refactor class menu (OB package Browser)

by Serge Stinckwich

2009/11/13 <csrabak(a)bol.com.br>: > Em 12/11/2009 19:29, Lukas Renggli <renggli(a)gmail.com> escreveu: > >> >> > That we don't Â have a way to create Â accessor (specifically the >> >> > getters) that do lazy evaluation? >> >> Lazy initialization? What do you want to do? >> > SomeClass>>aVariable >> > aVariable Â isNil ifTrue: Â [aVariable := Â "sensible thing Â to do"]. >> > Â Â ^aVariable >> Â How would the refactoring engine figure out what you want to do? > > We just give it some hints. Â To have an example of this functionality > pls get a look at the page nine of http://cormas.cirad.fr/forum/cormas/old2004_2006/att-0518/01-IntroCormas_El… Are you working with Cormas ? -- Serge Stinckwich UMI UMMISCO 209 (IRD/UPMC), Hanoi, Vietnam Smalltalkers do: [:it | All with: Class, (And love: it)] http://doesnotunderstand.org/

Nov. 13, 2009

Re: [Pharo-project] testing morphs

by Martin McClure

Tudor Girba wrote: > > This sounds interesting. Would it be possible to get a code sample (in > particular of the waitUntil: method)? > Sure. Here's the code we use in VW. I clearly didn't remember exactly what we called it. :-) We have a number of simpler methods that use this one, which is the most general case. Once you account for GemStone and VW-specific stuff, I imagine this approach should work in Pharo. Regards, -Martin gbtPauseUntil: validateBlock evaluatesReturning: expectedResult maxWaitSeconds: maxWaitSeconds "Evaluate validateBlock until it returns expectedResult or maxWaitSeconds have passed, pausing between evaluations. Return the last result of validateBlock value" | startTime result | startTime := Time secondClock. [result := validateBlock value = expectedResult] whileFalse: [Time secondClock - startTime > maxWaitSeconds ifTrue: [^result]. (GbxDelay forMilliseconds: 50) wait]. ^result

Nov. 13, 2009

Re: [Pharo-project] testing morphs

by Tudor Girba

Hi Martin, >> I would like to test the code related to rendering a morphic >> interface. In particular, I would like to test the rendering of >> Glamour browsers with morphic. >> >> The problem is that the rendering happens in another process than my >> regular test code. Any idea of how I could solve this problem? >> > > I'm not sure I understand exactly what you're wanting, but maybe you > do > some action in your main test code, this causes some rendering to > happen > in another thread, then after some unknown amount of time that thread > finishes rendering and comes to rest, and you want to test whether the > final state is what you expect? Indeed, this is what I want. > In that case the real problem is knowing when to expect the final > state > to be final. > One pattern we've used for UI testing is to have a method in the > test class > > waitUntil: conditionBlock > > You give this method a block that tests the condition and answers true > if all is good. The method runs the block every 20ms, using a Delay > between to allow the other thread(s) to run. As soon as the > condition is > true, it returns. If five seconds pass without the condition ever > becoming true, it reports a test failure. This lets you test the > results > of a unit of work in another thread without slowing your tests with > worst-case delays. This sounds interesting. Would it be possible to get a code sample (in particular of the waitUntil: method)? Cheers, Doru > Regards, > > -Martin > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project -- www.tudorgirba.com "What is more important: To be happy, or to make happy?"

Nov. 13, 2009

Re: [Pharo-project] testing morphs

by Martin McClure

Tudor Girba wrote: > Hi, > > I would like to test the code related to rendering a morphic > interface. In particular, I would like to test the rendering of > Glamour browsers with morphic. > > The problem is that the rendering happens in another process than my > regular test code. Any idea of how I could solve this problem? > I'm not sure I understand exactly what you're wanting, but maybe you do some action in your main test code, this causes some rendering to happen in another thread, then after some unknown amount of time that thread finishes rendering and comes to rest, and you want to test whether the final state is what you expect? In that case the real problem is knowing when to expect the final state to be final. One pattern we've used for UI testing is to have a method in the test class waitUntil: conditionBlock You give this method a block that tests the condition and answers true if all is good. The method runs the block every 20ms, using a Delay between to allow the other thread(s) to run. As soon as the condition is true, it returns. If five seconds pass without the condition ever becoming true, it reports a test failure. This lets you test the results of a unit of work in another thread without slowing your tests with worst-case delays. Regards, -Martin

Nov. 13, 2009

[Pharo-project] testing morphs

by Tudor Girba

Hi, I would like to test the code related to rendering a morphic interface. In particular, I would like to test the rendering of Glamour browsers with morphic. The problem is that the rendering happens in another process than my regular test code. Any idea of how I could solve this problem? Cheers, Doru -- www.tudorgirba.com "Don't give to get. Just give."

Nov. 13, 2009

[Pharo-project] MailMessage

by Schwab,Wilhelm K

Hello all, I am looking for a way to format text in email. Html comes to mind (yes, I hate those emails too, but it would work and the recipients don't know any better<g>), as does attaching a file with the data. I recently had to switch smtp servers, and for whatever reason, can no longer send html email. The server says it accepted it, but the mail is never delivered; send plain text through it, and it works. It could simply be that the server wants no part of html email, but it would be nice to repeat the test with another client, and/or try attachments as a solution to my problem. I found MailMessage, and could not figure out how to set the recipient. After some searching, I found 151 on Mantis; it is a five year old fix from Frank Shearer for adding recipients. If I had to guess, things are still broken, but I will try it and see what happens. Sending this now while I'm thinking about it. If anybody knows a way around my current problems, please don't feel that you would be cheating me out of the fun of solving it myself :) Bill

Nov. 13, 2009

Re: [Pharo-project] No replace menu entry in the tools of Pharo image?

by csrabak＠bol.com.br

Em 12/11/2009 19:47, Adrian Lienhard <adi(a)netstyle.ch> escreveu: > Hi Cesar > I don't miss it. How does it work? When you select 'replace' instead of 'find' you get a window with _two_ entry fields one for the string you want to replace and another for the replacement (OK I concede I'm used to VW ;-) > I also wonder where you found the menu item 'set search string'? If you click the yellow button in the OB Package Browser in the source pane you'll find it (Pharo1.0rc1 Latest update: #10495). > If there is enough interest from other people we can add the > replace. But we should take care not to get long menus > again. Thinking about it, I would remove 'find again (g)' from the > context menu. Or does anybody use it from the menu, rather than with > the keyboard shortcut? You made an interesting point: for experienced people (who already learned the keyboard shorcuts) probably the menus are less useful, but for newcomers or less frequent coders, the menus are the way to remember the functionality available. So I think if we want to remove too many items from the menus we have to put back a workspace filled with an introductory document which among other things has the list of keyboard shortcuts. Better even would be to have an updated quick reference card to render in PDF! Regards, -- Cesar Rabak

Nov. 12, 2009

[Pharo-project] HernÃ¡n Morales Durand is available for work

by HernÃ¡n Morales Durand

Just want to comment you that I'm available for work. I can relocate to any part of the world. If anyone is interested, please contact me privately. Best regards, HernÃ¡n

Nov. 12, 2009

Re: [Pharo-project] accessors entry in submenu of refactor class menu (OB package Browser)

by csrabak＠bol.com.br

Em 12/11/2009 19:29, Lukas Renggli <renggli(a)gmail.com> escreveu: > >> > That we don't have a way to create accessor (specifically the > >> > getters) that do lazy evaluation? > >> Lazy initialization? What do you want to do? > > SomeClass>>aVariable > > aVariable isNil ifTrue: [aVariable := "sensible thing to do"]. > > ^aVariable > How would the refactoring engine figure out what you want to do? We just give it some hints. To have an example of this functionality pls get a look at the page nine of http://cormas.cirad.fr/forum/cormas/old2004_2006/att-0518/01-IntroCormas_El… > Normally you just change the code directly in the changes browser. > I see, was my plan B, but I'm experiencing the bug I describe... > >> > And last but not least, if I attempt to edit a method in the > >> > new window "Changes: refactor class, accessors" I get an MNU: > >> > OR2AddMethodChangeNode(Object)>>doesNotUnderstand: > >> > #selectedClass. > >> I cannot reproduce the problem in the standard OmniBrowser. I > >> commonly edit the changes before applying the refactorings. > > This is happening in an image: Pharo1.0rc1 Latest update: #10495 > Looks like a bug in OCompletion. > I am still using ECompletion, so I've not seen that bug before. OK. This begs the question: which of the completions engines is supposed to be the 'official' one? Regards, -- Cesar Rabak

Nov. 12, 2009

Re: [Pharo-project] No replace menu entry in the tools of Pharo image?

by StÃ©phane Ducasse

> If there is enough interest from other people we can add the replace. > But we should take care not to get long menus again. Thinking about > it, I would remove 'find again (g)' from the context menu. Or does > anybody use it from the menu, rather than with the keyboard shortcut? > > Cheers, > Adrian For me I often want to do replace all. find should work like find again? Stef

Nov. 12, 2009