lkml.org 
[lkml]   [2009]   [Apr]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [GIT PULL] XFS update for 2.6.30
Date
On Apr 3, 2009, at 12:02 PM, Linus Torvalds wrote:

> On Fri, 3 Apr 2009, Linus Torvalds wrote:
>
>> On Thu, 2 Apr 2009, Felix Blyakher wrote:
>>>
>>> Were there any problems pulling from the xfs repository?
>>
>> Sorry, no - just too much email, too many trees to look at, too many
>> people to argue with.
>>
>> Pulled.
>
> Side note - I almost unpulled afterwards.

That was my concern, i.e. it's not pulled without explicit
NAK. I knew about your possible concerns.

> You've done several apparently totally useless pulls from my tree at
> random points.

Yes, I noticed that, and agree with all your points even
before you brought them up.
I already started talking to people to improve my process.
The reason the intermediate pulls from your tree were done
is to make sure that new xfs patches would not conflict
with some other changes already in the mainline. That was
part of the maintainer cheat sheet given to me, and I
didn't realize the side effects of it.
I probably can verify the possible conflicts without pushing
the merges into the repository and reset the working tree to
pre pull state.

At any rate, I'll find some way to manage that without
cluttering the history with the merges.
Any suggestions are welcome.

Thanks,
Felix

>
>
> Daily "keep up-to-date with Linus' tree" pulls are _strongly_
> discouraged
> (read: if this continues, I'll just stop pulling from you), because it
> makes the history totally unreadable after-the-fact. It has some
> direct
> technical downsides (it makes it much harder to run "git bisect" and
> see
> what is going on), but apart from those direct downsides it just
> makes it
> much harder for me - or anybody else who wants to get an overview of
> what
> happened - to visualize things when history is messy.
>
> Instead of having a clear nice line of development that says "this
> is what
> happened to XFS", those merges have basically mixed up all your
> changes
> with all the random _other_ changes in the tree.
>
> In other words, having those extra merges makes the graphical tools
> almost
> useless for getting some kind of "what happened" overview.
>
> I realize that an occasional back-merge may be required to resolve big
> conflicts early, but they really have to be pretty big and immediate
> for
> it to be a win.
>
> Linus



\
 
 \ /
  Last update: 2009-04-03 19:45    [W:0.606 / U:0.232 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site