|
|
|
Re: Source Control File Merging [message #470099 is a reply to message #470090] |
Tue, 06 February 2007 14:19 |
Kenn Hussey Messages: 1620 Registered: July 2009 |
Senior Member |
|
|
Ronan,
I suppose that if you are merging these things as text, you'll have these
issues even without stereotype applications. Have you considered using the
XMLResource.OPTION_LINE_WIDTH save option? If you specify 1 as the value,
for example, every feature value will be saved on a different line...
Kenn
"Ronan Mulvaney" <rmulvaney@curamsoftware.com> wrote in message
news:eq9hv6$286$1@utils.eclipse.org...
> Hi James,
>
> We believe it's mainly the formatting of the attributes in the stereotypes
> that are causing the bulk of the issues.
>
> Say we have a stereotype line like:
> <Ronan:my_package xmi:id="_CIHIwLFoEdukDeYgm9KOzw" attribute1="value1"
> attribute2="value2" base_Package="_CGnX0rFoEdukDeYgm9KOzw"/>
>
> If two users have happened to each change a separate attribute as part of
> each's separate work, they are going to hit a merge conflict due to the
> attributes for the stereotype all being on one line.
>
> The solution we have come up with here is to reload the xmi file after
> saving and format each attribute onto a separate lines thus hopefully
> easing merging (we hope) when they come to checking in their code.
>
> I probably misled you by saying that it is since moving to the latest UML2
> format, rather this is the first time we've started using this in anger
> and this problem probably would have existed in the earlier version too.
>
> I was interested in whether anyone had hit this in the field and resolved
> it in any other way.
>
> Thanks.
>
> Ronan
>
>> Hi Ronan,
>>
>> Can you describe the kinds of changes that you are having difficulties
>> merging. I'm not sure why there would be major issues merging files if
>> the stereotype applications are written to the bottom of the file rather
>> than somewhere in the middle.
>>
>>
>> Regards,
>>
>> - James.
>>
>>
>> "Ronan Mulvaney" <rmulvaney@curamsoftware.com> wrote in message
>> news:ept81o$f3i$1@utils.eclipse.org...
>>> Hi all,
>>>
>>> We are using ClearCase to source control model files that we are working
>>> on and since moving from Rose cat files to a format based on UML 2.0.2
>>> (written our own custom Editor) we are having terrible issues merging
>>> changes to the files.
>>>
>>> We believe this may be mainly due to the stereotypes being written to
>>> the bottom of the file compared to the previous releases and are just
>>> wondering whether anyone has already overcome this issue or have any
>>> tips on the best format for the files to aid team development.
>>>
>>> Thanks in advance.
>>>
>>> Ronan
>>
|
|
|
|
|
Re: Source Control File Merging [message #582301 is a reply to message #470090] |
Tue, 06 February 2007 14:19 |
Kenn Hussey Messages: 1620 Registered: July 2009 |
Senior Member |
|
|
Ronan,
I suppose that if you are merging these things as text, you'll have these
issues even without stereotype applications. Have you considered using the
XMLResource.OPTION_LINE_WIDTH save option? If you specify 1 as the value,
for example, every feature value will be saved on a different line...
Kenn
"Ronan Mulvaney" <rmulvaney@curamsoftware.com> wrote in message
news:eq9hv6$286$1@utils.eclipse.org...
> Hi James,
>
> We believe it's mainly the formatting of the attributes in the stereotypes
> that are causing the bulk of the issues.
>
> Say we have a stereotype line like:
> <Ronan:my_package xmi:id="_CIHIwLFoEdukDeYgm9KOzw" attribute1="value1"
> attribute2="value2" base_Package="_CGnX0rFoEdukDeYgm9KOzw"/>
>
> If two users have happened to each change a separate attribute as part of
> each's separate work, they are going to hit a merge conflict due to the
> attributes for the stereotype all being on one line.
>
> The solution we have come up with here is to reload the xmi file after
> saving and format each attribute onto a separate lines thus hopefully
> easing merging (we hope) when they come to checking in their code.
>
> I probably misled you by saying that it is since moving to the latest UML2
> format, rather this is the first time we've started using this in anger
> and this problem probably would have existed in the earlier version too.
>
> I was interested in whether anyone had hit this in the field and resolved
> it in any other way.
>
> Thanks.
>
> Ronan
>
>> Hi Ronan,
>>
>> Can you describe the kinds of changes that you are having difficulties
>> merging. I'm not sure why there would be major issues merging files if
>> the stereotype applications are written to the bottom of the file rather
>> than somewhere in the middle.
>>
>>
>> Regards,
>>
>> - James.
>>
>>
>> "Ronan Mulvaney" <rmulvaney@curamsoftware.com> wrote in message
>> news:ept81o$f3i$1@utils.eclipse.org...
>>> Hi all,
>>>
>>> We are using ClearCase to source control model files that we are working
>>> on and since moving from Rose cat files to a format based on UML 2.0.2
>>> (written our own custom Editor) we are having terrible issues merging
>>> changes to the files.
>>>
>>> We believe this may be mainly due to the stereotypes being written to
>>> the bottom of the file compared to the previous releases and are just
>>> wondering whether anyone has already overcome this issue or have any
>>> tips on the best format for the files to aid team development.
>>>
>>> Thanks in advance.
>>>
>>> Ronan
>>
|
|
|
Powered by
FUDForum. Page generated in 0.03340 seconds