# How to improve this kind of API? **URL:** \ **Category:** Ruby\ **Created:** [February 9, 2006, 8:43pm UTC](https://www.ruby-forum.com/t/how-to-improve-this-kind-of-api/65118 "2006-02-09T20:43:00Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![Lyle\_J](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/l/45deac/32.png) [@Lyle\_J](https://www.ruby-forum.com/u/Lyle_J)\ **Post date:** [February 9, 2006, 8:43pm UTC](https://www.ruby-forum.com/t/how-to-improve-this-kind-of-api/65118/1 "2006-02-09T20:43:00Z")
I’m looking to make an API change for some methods in FXRuby and would appreciate some suggestions about how to improve them. The pattern shows up in several places in the code, but it always boils down to some widget attribute that can have a combination of bit-flags. For example, the 4-way splitter window can be configured to expand only its top-left pane: splitter1.expanded = TOP\_LEFT or to expand both the top left and top right panes: splitter2.expanded = TOP\_LEFT | TOP\_RIGHT And to check whether, say, the BOTTOM\_LEFT pane is expanded: (splitter3.expanded & BOTTOM\_LEFT) != 0 Now, what I’m thinking of doing is replacing the symbolic constants with Ruby symbols, and doing something like this instead: splitter1.expand(:top\_left) splitter2.expand(:top\_left, :top\_right) and then having checks like: splitter3.expanded? :bottom\_left I think that in order to “turn off” one of those bits, I’m probably also going to need to add something like: splitter4.unexpand(:bottom\_right) Now, the question is, is there some existing pattern in Ruby that this reminds you of? In other words, what is the Ruby Way to handle this kind of setting? Thanks in advance for your suggestions! Lyle ---
**Author:** ![Lyle\_J](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/l/45deac/32.png) [@Lyle\_J](https://www.ruby-forum.com/u/Lyle_J)\ **Post date:** [February 9, 2006, 9:26pm UTC](https://www.ruby-forum.com/t/how-to-improve-this-kind-of-api/65118/2 "2006-02-09T21:26:24Z")
DÅ?a Å tvrtok 09 Február 2006 20:42 Lyle J. napísal: > splitter2.expanded = TOP\_LEFT | TOP\_RIGHT > > reminds you of? In other words, what is the Ruby Way to handle this kind of > setting? > > Thanks in advance for your suggestions! > > Lyle Looks fine to me. Personally I wish more GUI toolkits provided functions/methods like set\_flag() and clear\_flag() (or the even rarer explicit getters and setters for all applicable flags) that take flags that are to be changed instead having to futz around with unintuitive catch-all style flag fields and doing bitwise fandango all the time. Keep up the food work. David V. ---
**Author:** ![Lyle\_J](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/l/45deac/32.png) [@Lyle\_J](https://www.ruby-forum.com/u/Lyle_J)\ **Post date:** [February 9, 2006, 9:33pm UTC](https://www.ruby-forum.com/t/how-to-improve-this-kind-of-api/65118/3 "2006-02-09T21:33:24Z")
David V. wrote: > style flag fields and doing bitwise fandango all the time. Keep up the food > work. As much as we all love FXRuby, I sure hope Lyle has another source of food. 😉 Seriously, it’s great to hear that the FXRuby interface will be getting more ruby-like. I like Ara’s suggestions for easy configuration from YAML.load-ed hashes. ---
**Author:** ![Lyle\_J](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/l/45deac/32.png) [@Lyle\_J](https://www.ruby-forum.com/u/Lyle_J)\ **Post date:** [February 9, 2006, 10:11pm UTC](https://www.ruby-forum.com/t/how-to-improve-this-kind-of-api/65118/4 "2006-02-09T22:11:16Z")
On 2/9/06, Simon Kröger [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > Well, i haven’t used FXRuby very much, so i might miss the problem with > the current interface, but from what i see below i think this new > version doesn’t justify breaking old code. OK. > > For example, the 4-way > > splitter window can be configured to expand only its top-left pane: > > > > splitter1.expanded = TOP\_LEFT > > Seems perfectly fine. Ok :top\_left if more ruby’ish but also more > dangerous. Why do you consider it more dangerous to use a symbol in place of a symbolic constant? Is it that Ruby would “notice” that a constant’s name is erroneous (e.g. misspelling TOP\_LEFT as TOPLEFT), but wouldn’t notice if you made the same mistake for a symbol or string (e.g. misspelling “top\_left” as “topleft”)? > The following isn’t meant to be ‘the soultion’ but perhaps food for > thoughts: Thanks. I need to ponder this… ---
**Author:** ![Lyle\_J](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/l/45deac/32.png) [@Lyle\_J](https://www.ruby-forum.com/u/Lyle_J)\ **Post date:** [February 9, 2006, 9:43pm UTC](https://www.ruby-forum.com/t/how-to-improve-this-kind-of-api/65118/5 "2006-02-09T21:43:08Z")
Lyle J. wrote: > I’m looking to make an API change for some methods in FXRuby and would > appreciate some suggestions about how to improve them. The pattern shows up > in several places in the code, but it always boils down to some widget > attribute that can have a combination of bit-flags. Well, i haven’t used FXRuby very much, so i might miss the problem with the current interface, but from what i see below i think this new version doesn’t justify breaking old code. > For example, the 4-way > splitter window can be configured to expand only its top-left pane: > > splitter1.expanded = TOP\_LEFT Seems perfectly fine. Ok :top\_left if more ruby’ish but also more dangerous. > or to expand both the top left and top right panes: > > splitter2.expanded = TOP\_LEFT | TOP\_RIGHT Also fine. Perhaps define a TOP constant for those who don’t like binary operations. > And to check whether, say, the BOTTOM\_LEFT pane is expanded: > > (splitter3.expanded & BOTTOM\_LEFT) != 0 Ok, agreed, this isn’t very nice. > Now, what I’m thinking of doing is replacing the symbolic constants with > Ruby symbols, and doing something like this instead: > > splitter1.expand(:top\_left) > splitter2.expand(:top\_left, :top\_right) Well, i don’t like using 2 parameters when it is only one logically. > and then having checks like: > > splitter3.expanded? :bottom\_left I don’t like that either. > I think that in order to “turn off” one of those bits, I’m probably also > going to need to add something like: > > splitter4.unexpand(:bottom\_right) Oh, and i realy dislike this. > Now, the question is, is there some existing pattern in Ruby that this > reminds you of? In other words, what is the Ruby Way to handle this kind of > setting? ## The following isn’t meant to be ‘the soultion’ but perhaps food for thoughts: require ‘set’ class Symbol def method\_missing meth, \*param set = Set[self] set.send meth, param.to\_a if set.respond\_to?(meth) end end class Splitter attr\_accessor :expanded end TOP\_LEFT = :top\_left TOP\_RIGHT = :top\_right BOTTOM\_LEFT = :bottom\_right BOTTOM\_RIGHT = :bottom\_right splitter = Splitter.new splitter.expanded = TOP\_LEFT p splitter.expanded #=\> :top\_left splitter.expanded = TOP\_LEFT | TOP\_RIGHT p splitter.expanded #=\>\ p splitter.expanded.include?(BOTTOM\_LEFT) #=\> false p splitter.expanded.include?(TOP\_LEFT) #=\> true splitter.expanded = :bottom\_left | :top\_left p splitter.expanded #=\> #\ ## splitter.expanded.delete(:top\_left) p splitter.expanded #=\> #\ cheers Simon ---
**Author:** ![Lyle\_J](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/l/45deac/32.png) [@Lyle\_J](https://www.ruby-forum.com/u/Lyle_J)\ **Post date:** [February 9, 2006, 11:03pm UTC](https://www.ruby-forum.com/t/how-to-improve-this-kind-of-api/65118/6 "2006-02-09T23:03:01Z")
DÅ?a Å tvrtok 09 Február 2006 21:33 Joel VanderWerf napísal: > David V. wrote: > > > style flag fields and doing bitwise fandango all the time. Keep up the > > food work. > > As much as we all love FXRuby, I sure hope Lyle has another source of > food. 😉 I cnt saem to tyep tpday…