Our “coding” atoms are methods, not classes. Having method per file we can 
trace the history of changes with method detail, not class. 
Now… you are complaining in the wrong problem. 

FileTree meta information (who causes your merge problems) is needed *just* to 
keep monticello compatibility. 
That’s what is plain wrong and should be at least optional :)

I would happily integrate a “PlainFileTree” protocol (without monticello 
information) as an option, if someone provides such extension. 
Crap, I could make it myself if I have the time!… maybe next insomnia night :P

Esteban

> On 08 Jul 2015, at 14:51, Alexandre Bergel <alexandre.ber...@me.com> wrote:
> 
> This is frustrating….
> 
> I have followed the instruction given on 
> https://github.com/ThierryGoubier/GitFileTree-MergeDriver
> And I got:
> 
> ~/HackathonSattose2015> git merge master
> pathToGitFileTree-MergeDriver/merge --version .merge_file_WAu4Yh 
> .merge_file_WxznI2 .merge_file_ZmiIKE: pathToGitFileTree-MergeDriver/merge: 
> No such file or directory
> fatal: Failed to execute internal merge
> 
> Why do we need file tree? The simple fileout is not enough? Most of languages 
> have one file per class. I do not see why we need to be different on that. Or 
> maybe there is something obvious I am missing….
> 
> Maybe I will go back to Smalltalkhub afterall…
> 
> Alexandre
> 
> 
>> On Jul 8, 2015, at 2:21 PM, Otto Behrens <o...@finworks.biz> wrote:
>> 
>> We have a .gitignore file that contains:
>> version
>> methodProperties.json
>> 
>> So, we don't bother.
>> 
>> The side effects of this are:
>> Without methodProperties.json, you have the default author and
>> meaningless timestamp.
>> Without the version file, loading does not work properly. So we have a
>> script that generates version files for each package before we load
>> into the image. At first, this script counted versions for the package
>> and got all the info from the git repository. But this was too
>> expensive (because the file system became slow when traversing the git
>> repo to find the meta data). So now, we just generate a version file
>> with a fixed author, date & time now and a UUID & version number
>> derived from the SHA1 of the .package directory.
>> 
>> We get all the meta-information directly from the git repo, using git
>> tools. This can be better with GitFileTree.
>> 
>> So we really just use basics of Monticello & Metacello and do the rest
>> externally from the image. We need to explore the available tools
>> more.
>> 
>> On Wed, Jul 8, 2015 at 12:18 PM, Alexandre Bergel
>> <alexandre.ber...@me.com> wrote:
>>> Hi!
>>> 
>>> Do I still need to press the merge button in Monticello when I am working 
>>> on a filtree git repository?
>>> I understand that no since the merging has to be done by Git, and not by 
>>> monticello.
>>> 
>>> I tried to not do the merge in monticello, but I get some conflicts when 
>>> doing the git merge. For example:
>>> 
>>> CONFLICT (content): Merge conflict in 
>>> Hackathon.package/monticello.meta/version
>>> CONFLICT (content): Merge conflict in 
>>> Hackathon.package/HProject.class/methodProperties.json
>>> CONFLICT (content): Merge conflict in 
>>> Hackathon.package/HClass.class/methodProperties.json
>>> 
>>> Why do we need this methodProperties.json and version files? Shouldn’t git 
>>> handle this?
>>> 
>>> How should I merge these files? Any experience?
>>> 
>>> Cheers,
>>> Alexandre
>>> 
>>> --
>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
>>> Alexandre Bergel  http://www.bergel.eu
>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>>> 
>>> 
>>> 
>>> 
>> 
> 
> -- 
> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
> Alexandre Bergel  http://www.bergel.eu
> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
> 
> 
> 
> 


Reply via email to