Re: Time.utc! and Time.localtime!

On Dec 14, 2005, at 1:22 PM, Daniel Hobe wrote:
> On 12/14/05, Eric Hodel <[e-mail]> wrote:
>> On Dec 14, 2005, at 11:36 AM, Daniel Hobe wrote:
>>
>>> Writing a script yesterday I found out, much to my surprise, that  
>>> the
>>> Time instance methods utc and localtime modify the receiver and  
>>> don't
>>> end in !.  Is this something that could change in 1.9?  It seems  
>>> very
>>> un-rubyish.
>>
>> I consider it not quite a ! method because it only changes the
>> representation of the time, not the time itself.
>>
>> $ ruby -e 't1 = Time.now; t2 = t1.dup.gmtime; puts t1 == t2, t1, t2'
>> true
>> Wed Dec 14 12:23:21 PST 2005
>> Wed Dec 14 20:23:21 UTC 2005
>
> I disagree that they are not really ! methods.  It changes the offset
> from utc, which I'm assuming is in a instance var or something.
> Because you are changing the receiver, I think it a solid candidate
> for changing to end in !

Typically a method that modifies the receiver has a non-! method that  
does not.

Hash#reject and Hash#reject!
Array#uniq and Array#uniq!
String#sub and String#sub!

Looking through these classes I don't see a ! method that lacks a  
non-! method.

Time doesn't have a non-! equivalent, so you should know to check if  
it modifies the receiver.

When using Hash#delete_if I sometimes forget if it modifies the  
receiver or not, so I check the documentation or perform a test.

Also, what should the Time.now.utc!.utc! return? nil like 'x'.sub! 
('x', '').sub!('x', '') or the receiver like ['x'].map! { |e|  
e }.map! { |e| e }?  When possible, the ! methods return extra  
information about what they do.

> A little background on the reason it bit me: I was displaying the time
> to the user (in puts time.localtime) and then thought I was saving the
> utc times to MySQL which doesn't save timezone. This broke when I went
> to read it out again and assumed everything was UTC.
>
> I suppose it could be put down to me not RTFM, but I think the
> behavior is surprising enough to warrant a change.

How did you discover #localtime and #utc?

PS: The documentation clearly states it modifies the reciever.

-- 
Eric Hodel - [e-mail] - http://segment7.net
This implementation is HODEL-HASH-9600 compliant

http://trackmap.robotcoop.com