Re: [Gnu-arch-users] Re: named patches, patch order, patch queue manager

[Tom Lord]

    > From: Miles Bader <address@hidden>

    > The one thing that bothers me is namespace pollution -- sure I
    > can use naming conventions to avoid gross conflicts (e.g., call
    > all my `named patches' something like CAT--patch-NAME--0), but
    > having a flat namespace for all branch names might be a bit
    > annoying after a while, e.g., when looking at abrowse output,
    > you'd see every single named patch, regardless of whether
    > obsolete or not, and for some of the major linux hackers,
    > there'd be _many_ such patches.

It's not that hard to work with multiple commit-to archives at once
and, as time has passed, people have been providing little command
tweaks that make it even easier.    So, that gives you a two-level
namespace and, moreover, one that interacts nicely with units of
administrivia like mirroring.

(As you earlier observed, having all my prism-merge branches in 2003b
was a questionable idea -- when I next cycle archives I intend to
factor them out exactly this way.)


    > Perhaps it could be as simple as having an `hidden' (or `closed') flag
    > in archive branches/versions/something, so that `tla abrowse' and `tla
    > branches' wouldn't show them by default (say unless given given a
    > --hidden option); of course branch creation &c would still see them.
    > Then after a patch was merged into Linus's tree, you could just mark it
    > as hidden.

    > This might also be useful in general, since probably people have certain
    > branches/categories that are basically obsolete, but which following
    > arch practic, probably shouldn't be deleted from the archive.

Hmm.

-t