I am using darcs to keep track of some experimental changes to a tree
that is maintained in CVS.

I have one tree (A) in which I track the state of the CVS HEAD. This
source is also under darcs control. I have another darcs tree (B), which
is not known to CVS.

My approach has been to keep tree A up-to-date with HEAD, making no
changes of my own. As I update A from CVS, I darcs record the changes.

Meanwhile, back at tree B, I am working on some changes.  As CVS changes
accrue in HEAD and I record them in tree A, I pull to B from A,
resolving any conflicts as I go.

There is an extensive set of boring files, because I have tried to avoid
keeping more files than necessary in darcs. From time to time I would
realize that some set of files should/should not have been included in
the first place, so I made the necessary changes to boring and added or
removed the files. in tree A.

At one stage I removed a set of output files from A. It is quite
probable that these files had been modified during a build process on
tree B. When I tried to pull to B the patch that deleted these file and
their containing directory in A, the process spun for at least some
hours.

Deleting the subtree manually from B did not solve the problem;
reverting the tree in B, then pulling the patch did work.

The big problem with this was the lack of any feedback on the process.
Darcs just disappears for a long, long time. The new user (like me) is
at a loss. I would like to see Darcs support in NetBeans, for example,
but in order for that to happen, it must be possible to programatically
intercept and diagnose these problems.

I'm sure I'm not saying anything that hasn't been said here before. Love
the product.

Peter


_______________________________________________
darcs-users mailing list
[email protected]
http://www.abridgegame.org/mailman/listinfo/darcs-users

Reply via email to