[Glass] Git Bitbucket problematic

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

[Glass] Git Bitbucket problematic

GLASS mailing list
Ciao,

        this message is not direct to gemstone problematic,

        but anyone can help me to understand and redirect me to right reference.

        I work with Git Bitbucket repository to interchange code from Pharo to Gemstone.

        Now i have the same package  ( with the same UUID ):

        Name: ConfigurationOfDtrBase-DTR.13
        Author: DTR
        Time: 18 December 2014, 6:39:27.286829 pm
        UUID: 000ca25e-c28c-4170-9fe7-bde0855c92b0
        Ancestors: ConfigurationOfDtrBase-DTR.12


        But in Pharo ( where i use the git push command ) it report some data

        while in Gemstone ( where i update the data with git pull command ) it report other  data.

        ( Into gemstone it report the right Pharo data with also some hold class and methods removed from the package in Pharo )

        How it's possible ?  The UUID  package  displayed with Monticello is same in the two environment .

        Thanks for any considerations,


                Dario


_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass
Reply | Threaded
Open this post in threaded view
|

Re: [Glass] Git Bitbucket problematic

GLASS mailing list
Dario,

I am not quite sure what problem you are trying to report ... it sould
help me fiugre out what might be happening if you give a detailed
description of of the EXACT commands you take on each machine ...

I also don't know what you mean by "report some data" and "report other
data" ... details of the EXACT data and differences would help here ...

Dale

On 12/19/14 7:07 AM, Dario Trussardi via Glass wrote:

> Ciao,
>
> this message is not direct to gemstone problematic,
>
> but anyone can help me to understand and redirect me to right reference.
>
> I work with Git Bitbucket repository to interchange code from Pharo to Gemstone.
>
> Now i have the same package  ( with the same UUID ):
>
> Name: ConfigurationOfDtrBase-DTR.13
> Author: DTR
> Time: 18 December 2014, 6:39:27.286829 pm
> UUID: 000ca25e-c28c-4170-9fe7-bde0855c92b0
> Ancestors: ConfigurationOfDtrBase-DTR.12
>
>
> But in Pharo ( where i use the git push command ) it report some data
>
> while in Gemstone ( where i update the data with git pull command ) it report other  data.
>
> ( Into gemstone it report the right Pharo data with also some hold class and methods removed from the package in Pharo )
>
> How it's possible ?  The UUID  package  displayed with Monticello is same in the two environment .
>
> Thanks for any considerations,
>
>
> Dario
>
>
> _______________________________________________
> Glass mailing list
> [hidden email]
> http://lists.gemtalksystems.com/mailman/listinfo/glass

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass
Reply | Threaded
Open this post in threaded view
|

Re: [Glass] Git Bitbucket problematic

GLASS mailing list
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?

        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?

        If i right understand i need work directly in gemstone to remove the hold definitions.

        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.

        And if move class into another package  what the system do?


        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

> Dario,
>
> I am not quite sure what problem you are trying to report ... it sould help me fiugre out what might be happening if you give a detailed description of of the EXACT commands you take on each machine ...
>
> I also don't know what you mean by "report some data" and "report other data" ... details of the EXACT data and differences would help here ...
>
> Dale
>
> On 12/19/14 7:07 AM, Dario Trussardi via Glass wrote:
>> Ciao,
>>
>> this message is not direct to gemstone problematic,
>>
>> but anyone can help me to understand and redirect me to right reference.
>>
>> I work with Git Bitbucket repository to interchange code from Pharo to Gemstone.
>>
>> Now i have the same package  ( with the same UUID ):
>>
>> Name: ConfigurationOfDtrBase-DTR.13
>> Author: DTR
>> Time: 18 December 2014, 6:39:27.286829 pm
>> UUID: 000ca25e-c28c-4170-9fe7-bde0855c92b0
>> Ancestors: ConfigurationOfDtrBase-DTR.12
>>
>>
>> But in Pharo ( where i use the git push command ) it report some data
>>
>> while in Gemstone ( where i update the data with git pull command ) it report other  data.
>>
>> ( Into gemstone it report the right Pharo data with also some hold class and methods removed from the package in Pharo )
>>
>> How it's possible ?  The UUID  package  displayed with Monticello is same in the two environment .
>>
>> Thanks for any considerations,
>>
>>
>> Dario
>>
>>
>> _______________________________________________
>> Glass mailing list
>> [hidden email]
>> http://lists.gemtalksystems.com/mailman/listinfo/glass
>
> _______________________________________________
> Glass mailing list
> [hidden email]
> http://lists.gemtalksystems.com/mailman/listinfo/glass

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass
Reply | Threaded
Open this post in threaded view
|

Re: [Glass] Git Bitbucket problematic

GLASS mailing list

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
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass
Reply | Threaded
Open this post in threaded view
|

Re: [Glass] Git Bitbucket problematic

GLASS mailing list
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 ...

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
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass
Reply | Threaded
Open this post in threaded view
|

Re: [Glass] Git Bitbucket problematic

GLASS mailing list
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
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass


_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass
Reply | Threaded
Open this post in threaded view
|

Re: [Glass] Git Bitbucket problematic

GLASS mailing list

On 01/13/2015 06:21 AM, Dario Trussardi via Glass wrote:
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
Dario, this is the "smoking gun" ... `git rm DTRCassa.package/DTRCassaComponent.class/instance/flagSetup.st` should stage the delete for commit ...

When I am working with git, I typically sit in the root directory of the repository and do a `git add .` and then do `git commit` ... the `git add .` should stage changes, new files and deleted files for the commit ... you might need to do `git add -A .` in some versions of git ...

But the fact that your removals were not staged definitely explains why the deletions were not showing up in GemStone ...

Dale

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass