[darcs-users] [patch273] Haddock merge2FL and fastRemoveFL in Pat... (and 4 more)

Eric Kow bugs at darcs.net
Mon Jun 28 15:21:30 UTC 2010

Eric Kow <kowey at darcs.net> added the comment:

On Sun, Jun 27, 2010 at 13:29:26 +0200, Reinier Lamers wrote:
> Op woensdag 23 juni 2010 17:24 schreef Eric Kow:
> >  l1     l2    ((%     r3     r4))   l5.3   l6  (case iii!)
> Isn't that a case iv? l5.3 can't even commute past the right bubble because r3 
> is in the right bubble. Just checking if I understand...

Hmm, yes, I think you're right and I hope this was a mere think-o
(circumstantial evidence: case iv is the exciting case, hence the
exclamation mark)

An example of a case iii would be a subsequent l7.3
Bubbles, bubbles, everywhere

Quoting myself:
>> iii. Next, if a patch commutes past only the right bubble (stopping at the
>>      middle), then it gets absorbed into the middle bubble
>> iv. The interesting case is what happens when the patch does not even commute
>>     past the right bubble.  Basically we redraw the boundaries between middle
>>     and right: we do so by pushing p as far down the right bubble as we can.  Where
>>     we stop becomes the new middle/right boundary.
>> Following this example (using % to indicate the middle/right bubble boundary)
>> ((%))   l1     l2     r3     r4     l5.3   l6  (case ii)
>>  l1    ((%))   l2     r3     r4     l5.3   l6  (yawn)
>>  l1     l2    ((%))   r3     r4     l5.3   l6  (case i, r3 ADDED TO RIGHT)
>>  l1     l2    ((%     r3))   r4     l5.3   l6  (case i)
>>  l1     l2    ((%     r3     r4))   l5.3   l6  (case iii!)
>>  l1     l2   ((r3     l5.3 % r4))   l6         (case ii)
>>  l1     l2     l6   ((r3     l5.3 % r4))

Note that one confusing notational point is that the (case foo) step means
"what next" and not "how did we get here?"

Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9

Darcs bug tracker <bugs at darcs.net>

More information about the darcs-users mailing list