On 11 Feb 2000, Lars Gullik Bjønnes wrote:

> Allan Rae <[EMAIL PROTECTED]> writes:
> | When I instead use:
> |     cvs update -j HEAD
> 
> What does the cvs manual say about this?
> Perhaps HEAD is not a good tag to merge against.

The manual gives all its examples as merging a branch into the trunk but
none for one branch to another or for updating a branch.  I figured this
should mean I can do the reverse by reversing the order of the -j options.

> | I've got this compiling now (on a P100 this will take forever) and will
> | commit it sometime after it successfully works... maybe even tomorrow ;-)
> | 
> | I can certainly see why Lars does all his development in the main trunk.
> 
> Actually I do most of the development off-line in a checked out copy
> of the main trunk... and I (try to) only commit after I have finished
> the new module/feature/thingie ...

I'm not complaining about that.  It's just that you won't see most of
these problems because you don't suffer from a 3-way merge when updating.
That's why "cvs update -j HEAD" causes so many problems -- maybe I should
have tried "cvs update -j rae-1 -j HEAD" instead -- the three way merge
sees the original files that started the branch, my more up to date 
working directory and the HEAD sources.  Hmmm... actually that makes more
sense.  I might abandon this merge and wait for the updated painter
instead and then try again with "rae-1".

Allan. (ARRae)

Reply via email to