Re: [Gnu-arch-users] Broken trees?

[Frederic Gedin]

Jean Helou a écrit :

> Frederic Gedin a écrit :
>
> > Hi
> >
> >
> > I am facing a strange situation and I don't know how to fix it.  I 
> > develop either on my desktop or on my laptop.  On the desktop, I have 
> > an archive which contains the version "aida--main--0.1".  On the 
> > laptop, I have an archive which contains the version 
> > "aida--laptop--0.1".  Today I am unable to report changes made on the 
> > laptop into the desktop development tree.  The star-merge command 
> > gives me the following error message:
> > star-merge: unable to merge unrelated trees.
> >
> >  
>
> If I understand correctly, you say you have :
> aida--main--0.1, on your desktop
> aida--laptop--0.1, on your laptop tagged from aida--main--0.1
>
> and then you say you have a development tree on your desktop. If that 
> tree isn't aida--main--0.1, but instead is something like  :
>
> aida--anything--0.1 , tagged from aida--main--0.1 (wild guess) 

Actually no.  I develop directly in aida--main--0.1, on the desktop.

> Then the (current) star-merge algorithm won't detect that 
> aida--laptop--0.1 and aida--anything--0.1 do have a common ancestor, 
> because that ancestor is located in aida--main--0.1.
>
> I once had this problem too, two solutions :
> -make a replay/update from laptop to devo ( or devo to laptop doesn't 
> matter) so they have a common ancestor.
> -use the "-r, --reference VERSION" option to star-merge. 

I tried without any success.  However, you give me the right cause: the 
update comands on the desktop gives me the following message "update: 
tree has no common history with version".

The problem is unfortunately not symetric: on the laptop the star-merge 
with aida--main--0.1 works fine but on the desktop the star-merge with 
aida--laptop--0.1 doesn't. So apparently the laptop knows that 
aida--laptop--0.1 has aida--main--0.1 as an ancestor but the desktop 
doesn't.

So the question is, how can I set/check the common ancestor of two 
versions?

> Btw, I wonder how hard it would be to go past tag boundary to find a 
> common ancestor. Has anyone thought/tried to do something about it ? 

Sorry for not being mature enough on tla to help you there

Regards

Frédéric