-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Chris Nelson wrote:
> Alex Willmer wrote:
>> On Tue, Nov 9, 2010 at 9:39 PM, Chris Nelson
>> <[email protected]> wrote: 
>>> How is no_cycle different from parent_child?  I imagine parent_child
>>> has more than no_cycle.  It's only just occurred to me that cycles
>>> have to be detected along all possible links.  For example, A has a
>>> child B which blocks C which blocks A.   
>> no_cycle checks that the graph is acyclic like you say, preventing a
>> single link type forming a loop e.g. A depends on B depends on C
>> depends on A. parent_child in addition checks that a single link type
>> forms a tree - namely that a ticket may have many children, but at
>> most one parent.
> 
> That makes sense.  But is preventing "single link type" loops sufficient?  It 
> seems that my example could be a concern.

I feel like a collection of logic is needed here, what leads to a
two-stepped approach:
a) collect and somehow combine logic for every single relation type that
makes loop conditions possible

«strong» links like 'parent-child' or 'sibling' could be part of a loop,
while «weak» ones like 'relative' would not be concerned.

b) do a one-pass check with an algorithm respecting that combination of
possible loop paths


However, I don't know, if especially the combined logic could be done in
terms of 'doable within months, not years'.

Steffen Hoffmann
(hasienda)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEUEARECAAYFAkzcUPoACgkQ31DJeiZFuHcn+wCeJ/aPv0APFNi2afQql7tPoMXO
nGMAljRE0VjplWCLc84/cf7OB4Oqdo4=
=+hFU
-----END PGP SIGNATURE-----

-- 
You received this message because you are subscribed to the Google Groups "Trac 
Development" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/trac-dev?hl=en.

Reply via email to