About chronology....

ducasse ducasse at iam.unibe.ch
Sat Dec 13 10:14:50 UTC 2003


Another important question is the following one:

	do we want to keep track using for example monticello convention of 
the class extensions in String and other classes?
	my feeling is that this would be good to have *chronology-private 
*chronology-accessing....

Stef

On 13 déc. 03, at 10:59, ducasse wrote:

> hi all
>
> As you saw we are preparing the inclusion of the chronology package as 
> listed in the 3.7 alpha page on minnow.
> Please pay attention that the changes are not the one available on 
> squeak map but version1.3b2.
>
> So if you have to say something please say it now or ....
>
> Stef
>
> Here is the description brent sent to me abut beta3 (I sent the files 
> containing code, marcus will harvest the test separately).
>
>
>> The motivation behind the Chronology package was to forever remove the
>> problems of managing date/time sofware using Smalltalk's features (e.g
>> block closure iterations).
>>
>> I also wanted to use it as a demonstration of Smalltalk - indeed the
>> current julian date implementation of Date is one on my earlier 
>> changesets.
>> Hance non core classes like Schedule.
>>
>> As a basis the Chronology package implements both the ANSI standard 
>> and has
>> full backward comatibility (* see note) with the Smalltalk-80 
>> protocol and
>> supports some of the temporary stop-gap classes like TimeStamp.
>>
>> The inspiration, as I acknowledge on the Swiki, is not mine, someone 
>> else
>> first introduced explicit classes for Month, Week; I just took is to 
>> its
>> conclusion.
>>
>> All deprecated methods use the KCP latest deprecation mechanism.
>>
>> There are SUnit tests - one test that consistently fails is 
>> #testCoverage
>> which is merely my mechanism to determine which classes do not have 
>> tests
>> for all methods. I also use the ANSI tests on SqueakMap.
>>
>> All the significant tests are complete (alas I converted to XP after 
>> I has
>> written most of the code)
>>
>> The 1.3 beta introduces support for classes/methods used by Chris 
>> Muller's
>> Magma database. It is my intension to help him migrate to the 
>> chronology
>> package if possible (e.g class Stopwatch) as Chronology clashes with 
>> Magma.
>>
>> The SAR provides you with the option to just install the classes, or 
>> to
>> install the classes with the SUnit tests.
>>
>> (Embedded in the SAR is a copy the ANSI compatibility classes from
>> SqueakMap WITHOUT (sans) the DateAndTime and Duration classes, to make
>> testing ANSI compliance easy)
>>
>> IMPORTANT: The import may appear complex and daunting but it isn't 
>> really -
>> the complexity comes is trying to swap out the Date and Time classes 
>> whilst
>> they are being used by the file-in code. The rest is just 
>> factorisation.
>>
>> For the confident, the change sets in the SAR are named 
>> alphabetically in
>> the order they are to be imported. Some as mentioned are optional.
>>
>> Note: The only incompatibility is that the Week indexOfDay: #Sunday 
>> answer
>> 1.
>>
>> ToDo: Should my implementation make the grade:
>> 1. I think we should canvas Chris Muller about migrating Magma
>> 2. I need to categorise the methods properly.
>>
>> Tested in 3.7a-5595
>>
>> Thanks for your interest
>>
>> Brent
>>
>
>




More information about the Squeak-dev mailing list