Hello Forum,
We're nearly finished with the raw encapsulation of all of the Xtreme Suite's ActiveX controls, and we were wondering what Smalltalkers expect in the way of naming conventions. We ask because we intend to release our encapsulations into the community in case others wish to use this incredible product. For example, the XtremeCommandBars control has a method called ClosePopups(). The documentation reveals that a better name might be CloseAllPopups(). Doing things the Smalltalk way would, indeed, imply that our selector should be #closeAllPopups, the word "All" appearing to be a common feature in the Smalltalk idiom. For those of you with experience, would you find it an irritation if our naming conventions deviated from the control's naming conventions? Our thinking is that, once we've reached the stage of high-level encapsulation, the Smalltalk style should prevail. But then on the other hand, if one were trying to map the controls' documentation to our encapsulation, one might find deviations a hindrance. Any thoughts? Cheers, Eric |
Eric,
> For those of you with experience, would you find it an irritation if our > naming conventions deviated from the control's naming conventions? Our > thinking is that, once we've reached the stage of high-level > encapsulation, the Smalltalk style should prevail. But then on the > other hand, if one were trying to map the controls' documentation to our > encapsulation, one might find deviations a hindrance. You can approach this the same way OA treats COM: have a potentially private method for each virtual method and then add wrapper/helper methods as you see fit. With the control docs in hand, one can start from the "primitives" and browse for references to find the wrappers. You should ensure that anything you write yourself is not in the auto generated category. Have a good one, Bill -- Wilhelm K. Schwab, Ph.D. [hidden email] |
Bill,
We're pretty comfortable with the _process_ of wrapping. We're just not sure on the naming conventions. >...have a potentially private method for each virtual method and then add >wrapper/helper methods as you see fit. Hmmm. We'll have to look at this. I'm sure we'll be refining these encapsulations as we put them into practice. Thanks very much. Cheers, Eric > -----Original Message----- > From: Bill Schwab [mailto:[hidden email]] > Posted At: Wednesday, June 14, 2006 12:59 PM > Posted To: comp.lang.smalltalk.dolphin > Conversation: Naming Conventions in High-Level Encapsulations > Subject: Re: Naming Conventions in High-Level Encapsulations > > Eric, > > > For those of you with experience, would you find it an irritation if > > naming conventions deviated from the control's naming conventions? Our > > thinking is that, once we've reached the stage of high-level > > encapsulation, the Smalltalk style should prevail. But then on the > > other hand, if one were trying to map the controls' documentation to our > > encapsulation, one might find deviations a hindrance. > > You can approach this the same way OA treats COM: have a potentially > private method for each virtual method and then add wrapper/helper > methods as you see fit. With the control docs in hand, one can start > from the "primitives" and browse for references to find the wrappers. > You should ensure that anything you write yourself is not in the auto > generated category. > > Have a good one, > > Bill > > -- > Wilhelm K. Schwab, Ph.D. > [hidden email] |
"Eric Taylor" <[hidden email]> wrote in message
news:001301c68fee$c5f230d0$6500a8c0@server... > Bill, > > We're pretty comfortable with the _process_ of wrapping. We're just not > sure on the naming conventions. > >>...have a potentially private method for each virtual method and then > add >wrapper/helper methods as you see fit. > > Hmmm. We'll have to look at this. I'm sure we'll be refining these > encapsulations as we put them into practice. > > Thanks very much. > Bill is right; the Smalltalk style would be to rename the operations so that they make sense when spoken as English. The ability to quickly follow reference chains in the IDE is a core skill in Smalltalk development, so programmers should be well used to doing this and can easily use that to map from control documentation to your higher level abstractions. Lastly you'll find (no doubt have found ) that there are places in any abstraction where you want to deviate from the underlying control significantly, so to avoid the "analysis paralysis" of attempting to keep a mapping between the control documentation and your abstraction, I would just let go of the relationship entirely and not worry about it :-) Regards Blair |
Hello Blair,
>you'll find (no doubt have found) that there are places in any abstraction >where you want to deviate from the underlying control significantly... Indeed! That's what prompted our question :) Other languages are beginning to seem antiquated and unexpressive compared to Smalltalk. We just wanted to make sure that we could "let go," as you say. Cheers, Eric > -----Original Message----- > From: Blair McGlashan [mailto:[hidden email]] > Posted At: Wednesday, June 14, 2006 3:19 PM > Posted To: comp.lang.smalltalk.dolphin > Conversation: Naming Conventions in High-Level Encapsulations > Subject: Re: Naming Conventions in High-Level Encapsulations > > "Eric Taylor" <[hidden email]> wrote in message > news:001301c68fee$c5f230d0$6500a8c0@server... > > Bill, > > > > We're pretty comfortable with the _process_ of wrapping. We're just > > sure on the naming conventions. > > > >>...have a potentially private method for each virtual method and then > > add >wrapper/helper methods as you see fit. > > > > Hmmm. We'll have to look at this. I'm sure we'll be refining these > > encapsulations as we put them into practice. > > > > Thanks very much. > > > > Bill is right; the Smalltalk style would be to rename the operations so > that > they make sense when spoken as English. The ability to quickly follow > reference chains in the IDE is a core skill in Smalltalk development, so > programmers should be well used to doing this and can easily use that to > map > from control documentation to your higher level abstractions. Lastly > you'll > find (no doubt have found ) that there are places in any abstraction where > you want to deviate from the underlying control significantly, so to avoid > the "analysis paralysis" of attempting to keep a mapping between the > control > documentation and your abstraction, I would just let go of the > relationship > entirely and not worry about it :-) > > Regards > > Blair |
Free forum by Nabble | Edit this page |