[gentoo-dev] Re: Using cvs as your portage directory

[Michael Sterrett -Mr. Bones.-]

On Tue, 23 Mar 2004, Paul de Vrieze wrote:

> I might want to state that I personally do not use cvs as my portage tree. I
> have several reasons for this that you might consider:

I do use cvs as my portage tree (on one of my gentoo boxes) so I'll offer
some counter points for discussion.

> - It is always an advantage to see what the user sees

I agree with this (except for the "always"), but I've also found it useful
to see now what the users *will* see shortly.  Several times this has
enabled me to fix things that would have been noticed by the users.

> - syncing up the whole tree is both slow and it puts an extra strain on
>   cvs.gentoo.org

My usual routine is this: I sync the whole tree in the middle of the
night and then only update the parts that are touched (by watching
#gentoo-commits) during the day (if at all).  I'm asleep when the full
sync goes so my level of patience isn't an issue and the updates during
the day shouldn't put an unreasonable strain on the server.

> - It is very easy to accidentally commit temporary fixes you made in your life
>   tree. (this was esp. true with regard to package.mask that now can be
>   overridden without editing the life tree)

I agree with this point, but I think it's highly dependent on the style
of development that you use.  I have maybe one or two changes to the
portage tree that I'm testing on this box at any given moment so it's
easy to not commit those changes.  There are quite a number of ebuilds
in my local portage tree that are "in progress" but since they haven't
been added to CVS yet, there's no risk of committing them.  Generally,
I don't leave edits in the tree uncommitted.

Michael Sterrett
  -Mr. Bones.-
mr_bones_@g.o

--
gentoo-dev@g.o mailing list