[Glass] Git Bitbucket problematic

[Dario Trussardi via Glass]

Dale,


> Dario,
> 
> It occurs to me that if you are not correctly updating the git repository, then the proper information will not be transmitted to gemstone ...
> 
> When working in the git repository, you should use `git status` before and after each step, to verify that you have correctly accounted for the removed methods or classes ...

> 1. remove method from class

Into Pharo i remove the DTRCassaComponent 		flagSetup	 instance method.


> 2. save package to filetree/git repo

do it with monticello

> 3. in shell sitting in git repo on pharo machine: `git add 

submit: 	git add DTRcassa.package

Now the 	git status	 report:

# Changes not staged for commit:
#   (use "git add/rm <file>..." to update what will be committed)
#   (use "git checkout -- <file>..." to discard changes in working directory)
#
#	deleted:    DTRCassa.package/DTRCassaComponent.class/instance/flagSetup.st

If i right understand the deleted action is not initialized for commit, but because ?

The other :

	renamed:   
	new file:   
	modified: 

is right register to be committed.

Thanks for any considerations, 

	Dario

> Dale
> On 12/20/14 7:26 AM, Dale Henrichs wrote:
>> 
>> On 12/20/14 3:48 AM, Dario Trussardi via Glass wrote:
>>> Dale,
>>> 
>>>   i do some test for understund the continuos  development cycle from Pharo to Gemstone with git support.
>>> 
>>>   Question in my last email, i think is attributable at this problematic:
>>> 
>>>   if in Pharo i work on one class and remove or rename a method as this is management by git?
>> Yes.
>>> 
>>>   When i load the change into gemstone ( and this is not my first porting ) how the system management the remove method,
>>> 
>>>       and the rename method?
>> Loading from a git repository uses the same mechanisms as used when loading from an mcz file: method removal and renames should be handled correctly...
>>> 
>>>   If i right understand i need work directly in gemstone to remove the hold definitions.
>> This should not  be necessary ... there must be something in your workflow that is not quite correct ... or you are hitting an odd bug...
>>> 
>>>   My first think is to move ( into Pharo )  this method into a specific category ( deleted  renamed ) so as to have clear reference into gemstone.
>>> 
>>>   The same problematic is relative to the class definitions into package.
>>> 
>>>   If remove ( Or rename )  a class from a package, when i load it into gemstone, the hold class to be management manually.
>> Again, this should not be necessary for class renames either ... For class removal there are certain circumstances when GemStone will not remove a class: when the class has subclasses; but the Monticello loader usually get's it right ...
>>> 
>>>   And if move class into another package  what the system do?
>> If a class is move to another package, then the class may be able to survive, depending upon what the loaded package looks like
>>> 
>>> 
>>>   Ciao,
>>> 
>>>       Dario
>>> 
>>> P.S.  i select object database with objective to management continuous development with linear and clear definitions.
>>> 
>>>   I need to work some in this area
>> Again, I would like to have you described exactly what you do when you find that a method removed in pharo is not removed when the package is loaded in GemStone ... I assume that you have converted to using a BaselineOf to manage your packages? Here's the workflow you should be using:
>> 
>> 1. remove method from class
>> 2. save package to filetree/git repo
>> 3. in shell sitting in git repo on pharo machine: `git add .; git commit -m"..."; git push origin <branch name>`
>> 4. in shell sitting in git repo on gemstone machine: `git pull origin <branch name>`
>> 5. in gemstone:
>> Metacello new
>>   baseline: 'project-name';
>>   repository: 'filetree description';
>>   get;
>>   load;
>> 6. method should be remove from class in GemStone
>> 
>> Now if this sequence does not work, then we'll need to get more detailed and I will probably need a reproducible test case so I can figure out what's going wrong ...
>> 
>> Dale
>> 
>> 
> 
> _______________________________________________
> Glass mailing list
> Glass at lists.gemtalksystems.com
> http://lists.gemtalksystems.com/mailman/listinfo/glass

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gemtalksystems.com/mailman/private/glass/attachments/20150113/b089b4b9/attachment.html>