[MERGE] LockDir.lock_write() shouldn't require an email address

Martin Pool mbp at canonical.com
Wed Oct 4 02:11:04 BST 2006


On 04/10/2006, at 00:39 , Richard Wilbur wrote:

>> Because of inheritance, we don't need to do a deprecation phase,  
>> because
>> NoEmailInUsername is a child of BzrError.
>>
> John,
>
> I searched the bazaar-vcs.org site for "deprecation phase" and it
> returns with 0 pages/2344.  I searched for "deprecation" and it  
> returned
> five pages of which two mentioned bzr code.  BzrHistory under version
> 0.7 speaks of bzrlib.symbol_versioning being able to add is_deprecated
> flag to methods, change their doc strings, and cause them to issue
> DeprecationWarnings when called.
>
> Is this related to the "deprecation phase" you mention above  
> (albeit not
> needed in this case)?  Where could I read a description of this  
> practice
> ("deprecation phase")?

Hi Richard,

Good question.

Yes, "deprecation phase" just means "period when it's deprecated",  
nothing more.  Normally when we want to remove an interface we will  
make it deprecated for one or more release cycles, when using it will  
give a warning.  Then it's removed altogether.  John is saying it's  
not needed here.

Cheers
-- 
Martin







More information about the bazaar mailing list