Thread (6 messages) flat view 6 messages, 3 authors, 2016-06-15

Re: [TopGit] Multiple concurrent sets of patches

From: Jonas Smedegaard <hidden>
Date: 2016-06-15 22:46:19

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Tue, Mar 03, 2009 at 08:42:32PM +0100, Uwe Kleine-König wrote:
On Tue, Mar 03, 2009 at 08:22:21PM +0100, Jonas Smedegaard wrote:
quoted
On Tue, Mar 03, 2009 at 02:03:16PM +0100, martin f krafft wrote:
quoted
also sprach Jonas Smedegaard [off-list ref] [2009.03.03.1237 +0100]:
quoted
It seems to me that TopGit is incapable of handling this. That it 
can only handle patchset against a single branch, and if the need 
arise for restructuring an additional patchset for e.g. a stable 
or oldstable branch, then quilt needs to be used manually anyway.
Let me try to understand you: you want TopGit to maintain a single 
feature branch against two different source branches? How should 
that work? Could you elaborate a bit so that your needs become a bit 
more obvious?
Not quite. I want TopGit to maintain multiple feature branches, 
preferrably related.

With "related" I mean that I would like to be able to "fork" a chain 
of interdependent feature branches.

TopGit easily support one chain of branches:

upstream + pristine-tar -> master -> build

I want TopGit to additionally support the following:

upstream + pristine-tar -> stable-master -> stable-build

upstream + pristine-tar -> oldstable-master -> oldstable-build


E.g. in addition to TopGit branches t/fix_01 and t/feature_01 I would 
want to fork those branches as t_stable/fix_01 and t_stable/feature_01.


I know that I can create all those TopGit branches one by one, but I 
would then need to explicitly declare a list of TopGit branches to 
apply each time I want to (re)generate a quilt patchlist.
For my kernel development I use topgit topic branches that collect 
topgit patch branches.  Take this dependency graph:

linus/master <- t/armmisc/patch1 <- t/armmisc-master <- t/armmisc-pu
         ^ ^----t/armmisc/patch2 <--'                   |
                `------t/armmisc/patch3 <----------------------'

So t/armmisc-master collects patches that are ready for upstream, -pu
patches that need some more work.
My head is spinning - the dependency graph in my head was upside down 
compared to your ascii art here.

I think I understand it . Thanks!

I will try it out - I converted ghostscript last week to using TopGit as 
an excercise: good challenge, a nice bunch of patches (but nothing near 
as complex as kernel stuff, I know).

tg create t/fixes-squeeze/CVE-2009-abcd master-squeeze t/fixes-etch/CVE-2009-abcd
... no need to apply patch
... cherry-pick .topmsg from t/fixes-lenny/CVE-2009-abcd
git commit
Above puzzles me: Is cherry-picking .topmsg just a lazy way to write 
same description for parallel topicbranch, or is it a trick to cheat 
TopGit into believing that that branch never needs updating?

But this doesn't help me for my kernel problem, because here I don't 
have that ordering on releases.  I want to manage patches on top of 
linux-tip and the ARM tree, but none of these contains the other :-/
I can only begin to imagine the complexities of dealing with parallel 
tracks of kernel patching... Good luck with that!


Thanks a lot for your insight,

  - Jonas

- -- 
* Jonas Smedegaard - idealist og Internet-arkitekt
* Tlf.: +45 40843136  Website: http://dr.jones.dk/

  [x] quote me freely  [ ] ask before reusing  [ ] keep private
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkmtqg0ACgkQn7DbMsAkQLhMFACgkW1FYlWnEP8unk+wVgAkd0i+
UMwAoKT1mkxIRP6XmNKt+Rj3iJn2yUpM
=7x2Z
-----END PGP SIGNATURE-----
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help