Re: [BK] upgrade will be needed

["Sean" <seanlkml@sympatico.ca>]

On Fri, February 18, 2005 11:27 am, Theodore Ts'o said:

> If you truly believe that BK would be able to add the value that it
> does to the kernel development process by using some other SCM as the
> master SCM, with BK being "underneath", as you proposed earlier, then
> you do not understand why BK is fundamentally better than the current
> open source SCM systems that are out there.

BK already feeds patches out at the head, surely if it's as powerful as
you think, it could feed a free SCM too for your non-bk friends in the
community.

> And people *can* use the tools of their choice today.  They can use
> CVS, and diff+patch, and suffer with all of the limitations that those
> tools have today.  And for people who are doing stuff around the
> periphery, quilt is often really the best tool for them.

The situation could be improved for these other tools if there wasn't so
much BK zealotry from those that use it.

> If it's about the whole ***kernel*** development community, then it's
> pretty clear that the current system works quite well.  All of the
> complaints have been coming primarily from SCM hackers, it seems, and
> not people who truly need the power of more powerful than downloading
> the bk snapshots, using the CVS export tree, and in the case where
> they need to look at the changes in a single changeset bkbits.net.

There's no technical reason for this limitation.

> The "cost" of using BK seems to be primarily more theoretical, and
> ideological, than real.  It's always seems to be about someone
> kvetching that they want to use SVN and get finely grained changsets
> through SVN, and they can't.  But how often does that happen, and
> what's so painful of getting the finely grained changeset through
> bkbits.net?  Not very.  So at the end of the day, it finally boils
> down to being all about ideology, doesn't it?

No.  It's just that the cost isn't being paid by you, so you don't care.


Cheers,
Sean.