On Thu, Jan 29, 2009 at 08:41:08AM +0200, Damyan Ivanov wrote: > -=| Rene Mayorga, Wed, Jan 28, 2009 at 11:05:18PM -0600 |=- > > > > 1) Patch is Debian specific and there is no need to send it to upstream > > I agree that this would be grasp by looking at the patch name > > > 2) Patch needs to be send to upstream > > 3) Patch is already sent, and needs to be removed on the next release if is > > included. > > I think the policy here should be that patches not specific to Debian > must be sent upstream. So if the 1. rule is used, anything that does > not fall there is sent upstream, no need for special notion. Agree. The notation is just to get track if the patch was already sent. Having a thought, maybe we add a patch and upload the package and wait a few days before send it to upstream, or if the people who prepare/uploads the package does not send it for any reason, some else can do it > > Each point on this can have then a prefix on the name that tell to everyone the > > state of the patch > > like 00X_foo_bar.patch for option 1, > > 10X_bleh.patch for option 2 > > and 20X_baz.patch for option 3 > > Numbers suggest ordering, which may not be the case. debian_ or deb_ > or fix_paths_on_debian (or other name that makes it obvious this is > Debian-specific) is what I'd use. We can change number by `prefix' using char, I don't have a strong opinion on the name/prefix here. > The patch description helps here. I think we have some sort of > standard headers established lately: > > # Author: Name <email> > # Description: what the patch does and why > # Debian-Bug: http://bugs.debian.org > (after writing these the patch is ready to be sent upstream) > # Upstream-Bug: http://upstream.bug/url > (after sending) > > Of course, having Debian-only "flag" encoded in the name makes it more > visible, so it would be nice. I use 'RT: #NNNN` instead of `Upstream-Bug' > Hmm. Is the idea to track all the patches that are not sent upstream, > but should be? Perhaps "Upstream-Bug: not to be reported" or some such > can help grepping? Yes, and at some point have a review for all patches and if they where send it to upstream or not. Having a thought on the Description, the main concern here is that not all patches has one at this moment,( the naming schema is worst since none packages has it now :p) ) , IMHO using a naming scheme could be easy to remember and track then a `description tag' Cheers. -- Rene Mauricio Mayorga | jabber: rmayorga@jabber.org http://rmayorga.org | -------------------------------------------------- 08B6 58AB A691 DD56 C30B 8D37 8040 19FA A209 C305
Attachment:
signature.asc
Description: Digital signature