[ENH][3.6] DeprecationPreference-avi

Julian Fitzell julian at beta4.com
Tue Sep 16 22:25:31 UTC 2003


Well, I think I would argue that if they don't make the effort to notice 
early, they will certainly notice when the methods are removed and their 
users complain. :)  On the other hand, if we're having a preference 
anyway, I don't see any reason why it couldn't be a three-way preference 
if someone wants to write that code.

And yes, I think this may be a good thing to tuck in as one last fix 
before a final release.  We really should have caught it earlier of 
course: I kept thinking of bringing it up but was busy and figured 
someone else would complain eventually.

Julian

Doug Way wrote:
> 
> This seems like a generally good idea and is perhaps worth adding to 3.6 
> before it's released... I agree that the deprecation pop-ups could be a 
> big problem for non-developers.
> 
> I wonder if not showing any warnings even in the Transcript as the 
> default might be going too far, though.  Maybe the warnings should still 
> show up in the Transcript if you're not in fix-deprecation mode?  Or 
> maybe we could have a three-way preference with Transcript-warnings as 
> the default.  (There's an example of a radio-button preference under the 
> halos section of the prefs window.)
> 
> I'm just concerned that if there are no warnings at all, package 
> developers might not even notice that methods they rely on have been 
> deprecated.  Perhaps this isn't a big problem in practice, I'm not sure.
> 
> - Doug
> 
> 
> Julian Fitzell wrote:
> 
>> Ok, fine, I buy it. :)
>>
>> Julian
>>
>> Avi Bryant wrote:
>>
>>> Well, it seems more useful to pop up a debugger so that you can see the
>>> full context that the deprecated call was made in, and have access to 
>>> the
>>> source to change it right there.  Although I find the notifer popping up
>>> to be extremely annoying as a user, it's exactly what I would want as a
>>> developer trying to fix all my calls to deprecated methods.  Since I see
>>> these as two quite distinct modes, a preference makes lots of sense.  
>>> And
>>> that way when you're not in fix-deprecation mode, you don't get a whole
>>> bunch of messages cluttering up your Transcript either.
>>>
>>> On Tue, 16 Sep 2003, Julian Fitzell wrote:
>>>
>>>
>>>> I'm not convinced we really need a preference for this.  It adds more
>>>> complexity and I still think you'd end up getting annoyed at other
>>>> people's packages' deprecations while trying to solve your own (as
>>>> opposed to just ignoring those lines in a transcript).  I think I'd
>>>> still vote for just having transcript output, but this enhancement 
>>>> would
>>>> certainly be better than the status quo.
>>>>
>>>> avi at beta4.com wrote:
>>>>
>>>>> from preamble:
>>>>>
>>>>> "Change Set:        DeprecationPreference-avi
>>>>> Date:            16 September 2003
>>>>> Author:            Avi Bryant
>>>>>
>>>>> Adds a preference to control whether or not to show warnings for
>>>>> deprecated methods.  The preference is off by default."!
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------ 
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
> 
> 
> 




More information about the Squeak-dev mailing list