[darcs-users] "replace" patch type

Kevin Smith yarcs at qualitycode.com
Sun Mar 28 16:41:27 UTC 2004


David Roundy wrote:
> On Sat, Mar 27, 2004 at 09:01:51PM +0100, Marnix Klooster wrote:
>>What Kevin wants is just a 'replace now' patch.  Which is actually not
>>really different from a hunk patch, so I don't think it would really add
>>something.  Or would it?
> 
> But on the whole, I don't think this sort of a patch would be helpful.

I agree. A 'replace now' patch would be safer, but also loses much of 
the 'value' of the existing replace patch. I don't see a need for it.

A couple other clarifying points (really responding to David, but I'll 
put them in this message anyway:

When I said that replace patches are unbounded by time space, I meant 
that they can have unintended effects in my repo later, or in other 
repos (now).

I say "unbounded" because it is theoretically possible for someone to 
have a fork for ten years before finally merging in my replace patch. 
Unlikely, sure. But possible. The longer the delay, the more likely the 
replace patch will cause a problem.

And it's not just one fork. The whole point of a distributed RCS is that 
there can be hundreds of branches active at any time.

Now, I have not had much experience pulling/applying patches into my 
repo from other related repos. In fact, I have only done it once so far, 
but it was a somewhat jarring experience. I should probably describe it 
in a separate email. I short, it seemed to discourage me from reviewing 
the incoming patch before (or even after) it got applied.

But I seem to be alone in my dislike of the replace patch. Everyone else 
seems happy with it, so I'll stop complaining about it...again...for now.

Kevin




More information about the darcs-users mailing list