Smalltalk › Squeak › Squeak - Dev

#valueWithPossibleArgument: vs. #cull:

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

4 messages Options

marcel.taeumel

#valueWithPossibleArgument: vs. #cull:

Hi, there.

I think that all (6) sends in Trunk to #valueWithPossibleArgument: can be replaced with #cull:, which is more flexible.

Does anyone recall the story of #cull:? Maybe we should just implement #valueWithPossibleArgument: on MessageSend (and maybe Symbol) to make it more consistent?

For context, here is how I stumbled on it:

#(1 2 3) collect: #even. "Works."

1 ifNotNil: #even. "Works."

#(1) ifNotEmpty: #first. "Does not work."

Please note that this question is not about "#(1 2 3) collect: #even" versus "#(1 2 3) collect: [:ea | ea even]". ;-) This is more like a case of surprising inconsistency...

Best,

Marcel

Nicolas Cellier

Re: #valueWithPossibleArgument: vs. #cull:

I think that cull: is cross-dialect, maybe born in VW (Travis Griggs?)

Le lun. 20 mai 2019 à 17:38, Marcel Taeumel <[hidden email]> a écrit :

Hi, there.

I think that all (6) sends in Trunk to #valueWithPossibleArgument: can be replaced with #cull:, which is more flexible.

Does anyone recall the story of #cull:? Maybe we should just implement #valueWithPossibleArgument: on MessageSend (and maybe Symbol) to make it more consistent?

For context, here is how I stumbled on it:
#(1 2 3) collect: #even. "Works."
1 ifNotNil: #even. "Works."
#(1) ifNotEmpty: #first. "Does not work."

Please note that this question is not about "#(1 2 3) collect: #even" versus "#(1 2 3) collect: [:ea | ea even]". ;-) This is more like a case of surprising inconsistency...

Best,
Marcel

image.png (27K) Download Attachment

Chris Muller-3

Re: #valueWithPossibleArgument: vs. #cull:

In reply to this post by marcel.taeumel

Hi, there.

I think that all (6) sends in Trunk to #valueWithPossibleArgument: can be replaced with #cull:, which is more flexible.

It may be that these replacements can occur in trunk, but please don't let it lead to, "ah, now valueWithPossibleArgument: has no senders so we can deprecate it." ;-/

For arguments that accept a Collection, coercing a single object into a single-element collection is a useful pattern for improving API usability. #valueWithPossibleArgument: does that.

Does anyone recall the story of #cull:? Maybe we should just implement #valueWithPossibleArgument: on MessageSend (and maybe Symbol) to make it more consistent?

+1. We should add #valueWithPossibleArgument: to MessageSend (and maybe Symbol), and keep it on BlockClosure too. It is useful separately from #cull:[...] for certain situations.

Best,

Chris

For context, here is how I stumbled on it:
#(1 2 3) collect: #even. "Works."
1 ifNotNil: #even. "Works."
#(1) ifNotEmpty: #first. "Does not work."

Please note that this question is not about "#(1 2 3) collect: #even" versus "#(1 2 3) collect: [:ea | ea even]". ;-) This is more like a case of surprising inconsistency...

Best,
Marcel

Nicolas Cellier

Re: #valueWithPossibleArgument: vs. #cull:

Le lun. 20 mai 2019 à 21:19, Chris Muller <[hidden email]> a écrit :

Hi, there.

I think that all (6) sends in Trunk to #valueWithPossibleArgument: can be replaced with #cull:, which is more flexible.

It may be that these replacements can occur in trunk, but please don't let it lead to, "ah, now valueWithPossibleArgument: has no senders so we can deprecate it." ;-/

For arguments that accept a Collection, coercing a single object into a single-element collection is a useful pattern for improving API usability. #valueWithPossibleArgument: does that.

Does anyone recall the story of #cull:? Maybe we should just implement #valueWithPossibleArgument: on MessageSend (and maybe Symbol) to make it more consistent?

+1. We should add #valueWithPossibleArgument: to MessageSend (and maybe Symbol), and keep it on BlockClosure too. It is useful separately from #cull:[...] for certain situations.

Best,
Chris

valueWithPossibleArgument: does one thing more than cull:

it fill missing arguments with nil

I don't know if this is super useful... I find the feature questionable personnally.

Anyway, no sender is using this feature, so all senders could rather use cull:

valueWithPossibleArgs: has no equivalent because no cull message takes an Array of arguments.

valueWithPossibleArgs: is already understood by MessageSend while valueWithPossibleArgument: is not which is somehow surprising.

I'm not sure at all that these two selectors would be good to add to Symbol.

For example, Symbol only understand cull: but not cull:cull: nor does it understand valueWithArguments:

For context, here is how I stumbled on it:
#(1 2 3) collect: #even. "Works."
1 ifNotNil: #even. "Works."
#(1) ifNotEmpty: #first. "Does not work."

Please note that this question is not about "#(1 2 3) collect: #even" versus "#(1 2 3) collect: [:ea | ea even]". ;-) This is more like a case of surprising inconsistency...

Best,
Marcel