Repeated merge issue

[Christian Boos]

Thomas Arendsen Hein wrote:
> * Christian Boos <cboos at neuf.fr> [20051216 22:56]:
>   
>> Not really... I spent some time to create a test script.
>> I hope that this will demonstrate my problem better.
>>     
>
> Yes, that helped me understand what we do differently, though it had
> some problems with changes done in the same second without changing
> the size, hg commit doesn't notice the changes then.
>
> Attached is a script which demonstrates what you want.
>
>   

Thanks. I understand that this could be an alternative approach,
though not as intuitive (in my point of view) as the update -m
approach.
In your approach, each discarded changeset will be stored
as a full diff (basically the reverse diff of the changeset itself),
whereas with mine (if it would have worked), this would have
been an empty diff, simply recording the merge of the two
parents.

Also, in my scenario/script, I don't think that hg had a problem
to notice the change because of a "same second/same size" issue,
because you will have the same outcome if you do the steps manually.

I have now a few days off the net, I'll investigate further why my
script didn't work as I expected.

-- Christian