Hi Marc,
> On 22 May 2015, at 18:43, Marc Herbert <[email protected]> wrote:
> stgit series are attached to a specific branch. I'm curious to know if this
> is an intended design choice or just a limitation of the current
> implementation.
It is intended design choice. When I first announced StGit (long time ago,
about two-three months after Git was announced), I didn't have branches in
mind, I just wanted a Quilt emulation on top of Git independent of which branch
is checked out. But the feedback at the time was that I really need to make a
series per branch, and it makes sense for (lots of) people who use a small
number of repository clones but many (topic) branches.
> For prototyping/experimenting/context switching, plain git actively supports
> proliferation of short and short-lived branches. In such a case most of my
> stgit patches are useful on most branches, and moving/exporting/importing
> stgit patches really gets in the way. Moreover, duplication makes the
> maintenance of the patches themselves become very difficult - very hard to
> remember which way to stg sync and when.
It's funny that I sometime have such need as well and looked into improving
"git stash" to be able to "pop" patches from a git branch. But I never got
around to finishing it.
StGit doesn't have the concept of "global" patches but you can use a single
branch for all your patches and keep rebasing it on top of whichever base you
want.
> Could this be a side effect of the "[RFC] StGit UI changes" in 2011?
> https://www.mail-archive.com/stgit-users%40gna.org/index.html#00012
Not really. The per-branch series happened before this time, I think even
before I started tracking StGit changes in Git (I'm away from a computer now to
be able to easily check).
The above was more about permanently committing patches to a Git branch.
> Is anyone else feeling the same or am I just too demanding trying to use both
> git and Stacked Git at the same time? Is there some magic stg command /
> workflow / alternative to Stacked Git that I missed?
I think it's more about the workflow but things can always be improved:
a) topic branches with StGit series per branch
vs.
b) single StGit series rebased onto different topic branches
Option (b) doesn't come without problems. For example, a patch merged into a
topic branch (or upstream) may cause a patch to be removed on rebase (or
changes). So it needs some care (we at least have "undo").
Similarly if we would have the concept global patches, a patch can change its
content depending on which branch it is pushed to. So in general I think this
would only work for patches never meant for upstream (e.g. I have a "debug"
branch with various hacks and when I need to investigate something I just
switch to it and rebase it into the topic branch I want to debug).
> PS: is there a decent, searchable mail archive for this list? It seem present
> on none of gmane, marc.info, Google groups,...
Not really, whatever gna.org provides. But it's not that much history on this
list as early discussions happened on LKML or the Git mailing lists.
Catalin
_______________________________________________
stgit-users mailing list
[email protected]
https://mail.gna.org/listinfo/stgit-users
