[Gnu-arch-users] microbranching + automatic cacherevs

[Johannes Berg]

Hi,

Here's an idea -- I just want to see if anyone thinks its feasible. And
I also want to spend some time thinking about it, so I'll just write it
up here ;-)
I often do a lot of branches from external projects like tla or moin.
Now, every time I tag, I get a new cacherev which can be pretty big (and
they accumulate). It has been pointed out to me though that not using
them is bad.

So, what I'm thinking is this. A hypothetical "aba tag" (if I write this
as a script it'd surely for aba) could do this:
[assuming for the sake of the example that I'm working on moinmoin; I'll
add faked patchlevels]

when I tag moin--main--1.2--patch-266 to moin--something--1.2 it could
check for the existence of moin--UPSTREAM-PATCH-266-main--1.2 in my
archive, and if that doesn't exist tag there first (cacherev done) and
then tag where I want (no cacherev done).

That works, but is not very satisfying. When I tag the next version for
my next branch, I get a new cacherev. Assuming I only move along with
development and never tag something old, I could also have:
UPSTREAM--moin--main--1.2--patch-1 == real patch-266 from upstream, then
when I a few days later tag again and its patch-268 then I get a
UPSTREAM--moin--main--1.2--patch-2 which just pulls in -267 and -268
from upstream into that UPSTREAM category...

I don't think it has to cross branches or version numbers to find some
ancestry though.

Thats all very nice, but it somewhat pollutes my patchlogs. I don't
really want to know about the UPSTREAM-xxx category in the patchlogs.
But then tla would need  to support something like this inside.

Maybe there should be cacherev support somewhere other than inside the
patch directories? Then at least I wouldn't cacherev the same identical
version 10 times.

Oh, and before you say this: Yes, I know, I can do this manually. But
even if I value used space on my hd, I also value my time ;-)

This is just some unsorted thoughts, maybe someone has a better idea? I
just feel that the cacherev is
a) necessary, but 
b) tends to get large
and I want an automatic solution for that ;-)

johannes
-- 
http://www.sipsolutions.de/
Key-ID: 9AB78CA5 Johannes Martin Berg <address@hidden>

signature.asc
Description: This is a digitally signed message part