[e-lang] Documentation bug for FlexList

[Kevin Reid]

On Mar 16, 2007, at 15:24, David Hopwood wrote:

> David Hopwood wrote:
>> Kevin Reid wrote:
>>> On Mar 15, 2007, at 23:43, David Hopwood wrote:
>>>
>>>> <http://erights.org/javadoc/org/erights/e/elib/tables/ 
>>>> FlexList.html>
>>>>
>>>> # void setRun(int start, int bound, EList other)
>>>> # Enabled: Replace from start..!bound in this list with other.
>>>>
>>>> ? def a := [2, 3, 4].diverge()
>>>> # value: [2, 3, 4].diverge()
>>>>
>>>> ? a.setRun(0, 0, [1])
>>>> ? a
>>>> # value: [1, 2, 3, 4].diverge()
>>>>
>>>> 1. Why is this called 'setRun'? 'insertRun' would make more sense.
>>>
>>> Note that the name "setRun" is part of the assignment sugar:
>>>
>>>   a(b, c) := d
>>>
>>> expands to
>>>
>>>   a.setRun(b, c, def ares__1 := d)
>>>   ares__1
>>>
>>> , and a(b, c) is of course a.run(b, c).
>
> Oh, you mean that "Run" refers to the verb "to run". I had interpreted
> it as the noun "run" meaning a sublist, e.g. a run of cards. That's
> English for you :-)

I hadn't intended to say that in particular. E does make that pun,  
however; "run" is used for both "running" a subroutine (function) and  
taking a "run" (subsequence) of a collection.

But my point was that there is a sugar which expands to this  
operation on lists, and that either the sugar must be changed to  
match, the interface be duplicated, or the sugar removed (or no  
longer be usable for this purpose).


Personally, I do not find the notion of "replacing" a subsequence  
with a different-length subsequence at all odd, so I have no problem  
with the replace/5 name. I wouldn't mind changing setRun/3 to  
replaceRun/3 if we agree that it's an improvement and that it works  
with non-list objects (or to discard the sugar connection);  
"insertRun" I find to be wrong in the case where the specified range  
is not zero-length.

-- 
Kevin Reid                            <http://homepage.mac.com/kpreid/>