[Kde-scm-interest] How to handle conversion errors

[Eike Hein]

On 02/28/2010 12:17 AM, Andreas Pakulat wrote:
> On 28.02.10 00:07:38, Oswald Buddenhagen wrote:
>> On Sat, Feb 27, 2010 at 11:56:39PM +0100, Andreas Pakulat wrote:
>>> On 27.02.10 13:09:13, Ian Monroe wrote:
>>>> 2010/2/27 Thiago Macieira <thiago at kde.org>:
>>>>> Without an example, we can't tell. And you can simply delete useless things at
>>>>> the end of the import.
>>>>
>>>> Just thought I'd add that if you do something like this, to be sure to
>>>> documented via a script file or something uploaded to kde-ruleset so
>>>> that we can have "The Switch" done in one weekend.
>>>
>>> I'm not sure I follow you 100%, but the rules I'm writing right now are
>>> in my local clone of kde-ruleset and once I'm a member of the
>>> kde-developers group I'll certainly push that.
>>
>>> However deletion of useless stuff will probably be manually at the
>>> end.
>>>
>> ian meant that you should automate that.
> 
> I can't see a way of automating that easily at the moment, i.e. to judge
> wether a tag or branch with _<somenumber> is needed or not. Same thing
> goes for a branch that ends up in nirwana (i.e. the first commit in the
> branch has no parent) or tags that have no parent.

With "automating" he meant including a script with the
rules that would allow someone else to do it again. Ian
was thinking you're helping to prepare the rules for
The Switch, not going to switch to git ahead of time.

BTW, tags pointing to orphaned commits tend to be a
sign of errors in the rules. E.g. old tags created by
cvs2svn need special handling with recurse rule, see
konversation-rules for an example.


-- 
Best regards,
Eike Hein