Oh,I am quite at fault. Thanks. 在 2016年8月26日星期五 UTC+8下午3:36:33，Jan写道： > > Hi, > > I think the results of > > FOR sv IN subject FILTER sv.name == "C3" FOR v,e,p IN 2 ANY sv > r_parent RETURN p > > should be identical to the results of > > FOR sv IN subject FILTER sv.name == "C3" FOR v,e,p IN 2 ANY sv > INBOUND r_parent OUTBOUND r_parent RETURN p > > apart from order of results inside the result set. > This also assumes that there are no edges in r_parent that have identical > values in _from and _to in the same edge (self-referencing edges). > > Can you provide some example data for which the query just using the > collection once with "any" produces too many results? > > Best regards > Jan > > > > Am Freitag, 26. August 2016 09:16:18 UTC+2 schrieb Beansprouts: >> >> Sorry,I made a misstake on the first part of my aql. >> if i just use an "any" direction,the set of results will be bigger than >> that i need. >> >> >> <https://lh3.googleusercontent.com/-Zf4ycDgNqVM/V7_svc65mgI/AAAAAAAAAAc/4GZ5kUoJINMi_OwjBSHy75oeMXXPYMC-ACLcB/s1600/123.png> >> >> >> 在 2016年8月26日星期五 UTC+8下午3:04:03，Jan写道： >>> >>> Hi, >>> >>> I have tried the query and got some parse errors in the part before the >>> traversal. >>> I think the first part should read >>> >>> FOR sv IN subject FILTER sv.name == "C3" >>> >>> (filtering on "sv" and not "v", and using a double equal sign to make >>> the expression a comparison and not an assignment). >>> >>> Apart from that, the traversal part specifies "sv" as the edge >>> collection and "any" as the traversal direction. Then it specifies >>> "r_parent" for outbound traversal and "r_parent" again for inbound >>> traversal. >>> Though the combination of the "inbound" and "outbound" directions is the >>> same as "any", I think the traversal part can be simplified to just >>> >>> FOR v,e,p IN 2 ANY sv r_parent RETURN p >>> >>> The query parser just checks for whether multiple directions are >>> specified for the same collection and will report an error if this is so. >>> >>> Best regards >>> Jan >>> >>> Am Freitag, 26. August 2016 08:53:40 UTC+2 schrieb Beansprouts: >>>> >>>> >>>> <https://lh3.googleusercontent.com/-614Soa26kRI/V7_jusf7VJI/AAAAAAAAAAM/PVmoJfFy7F48NOb-u-fVGNQ1hwX9MTxngCLcB/s1600/123.png> >>>> >>>> This is a small part of my graph,every vertex may has a parent and >>>> children. >>>> >>>> my case : we have a start vertex,then we need to find other verties >>>> have the same parent with the start vertex. >>>> >>>> In old version,we can do this by an aql: for sv in subject filter >>>> v.name="C3" for v,e,p in 2 any sv outbound r_parent,inbound r_parent >>>> return p >>>> >>>> But,in version 3.0.4 the same aql cannot work as before,just got a >>>> exception message "conflicting directions specified for collection". >>>> >>>> It's a new feature! >>>> >>>> But,I think it's an bug. >>>> >>>>
-- You received this message because you are subscribed to the Google Groups "ArangoDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
