cannot change edge's in/out with @rid

[tolgaulas]

OrientDB Version: 3.0.25

Java Version:

OS: Win 2019

Expected behavior

--before--
v1(#1:1) -out- edge1(#2:1) -in->v2(#1:2)
v1(#1:1) -out- edge2(#2:2) -in->v3(#1:3)

--op--
update edge #2:2 in=#1:2

--result--
v1(#1:1) -out- edge1(#2:1) -in->v2(#1:2)
v1(#1:1) -out- edge2(#2:2) -in->v2(#1:2)

Actual behavior

v1(#1:1) -out- edge1(#2:1) -in->v2(#1:2)
v1(#1:1) -out- edge2(#2:2) -in->v3(#1:3)

--op--
update edge #2:2 in=#1:2

--result--
v1(#1:1).out =[]
v2(#1:2).in =[#2:1,#2:2]
v3(#1:3).in=[]
edge2.in = #1:1
edge2.out = #1:2

Steps to reproduce

When I change an edge’s in/out property with @Rid (ie. update edge theEdge in=#12:12) the "out"ing vertex is loosing the link. And the simple traversing queries stop working.
Instead, one have to use the select notation instead of direct rid (i.e update edge theEdge in=(select from #12:12).

Yet, the "out"ing vertex still keeps an empty out_theEdge=[ ], which looks like having no impact in the internal engine, yet it’s disturbing.

[luigidellaquila]

Hi @tolgaulas

Thank you for reporting, I'll check it ASAP and let you know

Thanks

Luigi

[luigidellaquila]

Hi @tolgaulas

I just tried it with the following

create vertex V set name ='a';  /* #9:0 */
create vertex V set name ='b'; /* #10:0 */
create vertex V set name ='c'; /* #11:0 */

create edge E from #9:0 to #10:0  /* #21:0 */

update edge #21:0 set in=#11:0

And it seems to work fine.
Any chance to have a reproducer?

Thanks

Luigi

[tolgaulas]

Hi, can you run this and see select from V at the end in a blank DB please?

BEGIN;
LET a = DELETE VERTEX v1;
LET a = DELETE VERTEX v2;
LET a = DELETE VERTEX v3;
LET v1_r1 =INSERT INTO v1 set r=1;
LET v1_r2 =INSERT INTO v1 set r=2;

LET v2_r1 =INSERT INTO v2 set r=1;
LET v2_r2 =INSERT INTO v2 set r=2;

LET v3_r1 =INSERT INTO v3 set r=1;
LET v3_r2 =INSERT INTO v3 set r=2;

LET onetwo_oneone = CREATE EDGE one_two from $v1_r1 to $v2_r1;
LET onetwo_twotwo = CREATE EDGE one_two from $v1_r2 to $v2_r2;
LET onetwo_onetwo = CREATE EDGE one_two from $v1_r1 to $v2_r2;
LET onetwo_twoone = CREATE EDGE one_two from $v1_r2 to $v2_r1;

LET twothree_oneone = CREATE EDGE two_three from $v2_r1 to $v3_r1;
LET twothree_twotwo = CREATE EDGE two_three from $v2_r2 to $v3_r2;
LET twothree_onetwo = CREATE EDGE two_three from $v2_r1 to $v3_r2;
LET twothree_twoone = CREATE EDGE two_three from $v2_r2 to $v3_r1;

LET touchOne = update edge E set in=$v2_r1.@rid where @rid in (select unionall(inE()) from $v2_r2.@rid);

COMMIT RETRY 100;

RETURN;

[tolgaulas]

I made it .toJson() and here is the result: Mind the #60:2.

{"@type":"d","@rid":"#47:2","@version":1,"@class":"v1","r":1,"out_one_two":["#85:2","#87:2"],"@fieldTypes":"out_one_two=g"}
--
{"@type":"d","@rid":"#48:2","@version":1,"@class":"v1","r":2,"out_one_two":["#86:2","#88:2"],"@fieldTypes":"out_one_two=g"}
{"@type":"d","@rid":"#59:2","@version":1,"@class":"v2","r":1,"in_one_two":["#85:2","#88:2","#86:2","#87:2"],"out_two_three":["#97:2","#99:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}
{"@type":"d","@rid":"#60:2","@version":1,"@class":"v2","r":2,"in_one_two":[],"out_two_three":["#98:2","#100:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}
{"@type":"d","@rid":"#71:2","@version":1,"@class":"v3","r":1,"in_two_three":["#97:2","#100:2"],"@fieldTypes":"in_two_three=g"}
{"@type":"d","@rid":"#72:2","@version":1,"@class":"v3","r":2,"in_two_three":["#98:2","#99:2"],"@fieldTypes":"in_two_three=g"}

[tolgaulas]

And I guess, here may be problem:

select *.toJson() from V where in_one_two is not null

#60.2 should not be appeared here.

{"@type":"d","@rid":"#59:2","@version":1,"@class":"v2","r":1,"in_one_two":["#85:2","#88:2","#86:2","#87:2"],"out_two_three":["#97:2","#99:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}
--
{"@type":"d","@rid":"#60:2","@version":1,"@class":"v2","r":2,"in_one_two":[],"out_two_three":["#98:2","#100:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}

[luigidellaquila]

Hi @tolgaulas

Thank you very much.
Ok, so the problem is in a batch script, let me check it, I'll let you know ASAP

Thanks

Luigi

[tolgaulas]

Not necesarely in batch, here is the before state, the select op and the after state. See #62:2

== BEFORE== 
@version | *.toJson()
-- | --
0 | {"@type":"d","@rid":"#49:1","@version":1,"@class":"v1","r":1,"out_one_two":["#89:2","#91:2"],"@fieldTypes":"out_one_two=g"}
0 | {"@type":"d","@rid":"#50:1","@version":1,"@class":"v1","r":2,"out_one_two":["#90:2","#92:2"],"@fieldTypes":"out_one_two=g"}
0 | {"@type":"d","@rid":"#61:2","@version":1,"@class":"v2","r":1,"in_one_two":["#89:2","#92:2"],"out_two_three":["#101:2","#103:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}
0 | {"@type":"d","@rid":"#62:2","@version":1,"@class":"v2","r":2,"in_one_two":["#90:2","#91:2"],"out_two_three":["#102:2","#104:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}
0 | {"@type":"d","@rid":"#73:2","@version":1,"@class":"v3","r":1,"in_two_three":["#101:2","#104:2"],"@fieldTypes":"in_two_three=g"}
0 | {"@type":"d","@rid":"#74:2","@version":1,"@class":"v3","r":2,"in_two_three":["#102:2","#103:2"],"@fieldTypes":"in_two_three=g"}

==OPERATION==
update edge E set in=#61:2 where @rid in (select unionall(inE()) from #62:2);

==RESULT === 

0 | {"@type":"d","@rid":"#49:1","@version":1,"@class":"v1","r":1,"out_one_two":["#89:2","#91:2"],"@fieldTypes":"out_one_two=g"}
-- | --
0 | {"@type":"d","@rid":"#50:1","@version":1,"@class":"v1","r":2,"out_one_two":["#90:2","#92:2"],"@fieldTypes":"out_one_two=g"}
0 | {"@type":"d","@rid":"#61:2","@version":3,"@class":"v2","in_one_two":["#89:2","#92:2","#90:2","#91:2"],"r":1,"out_two_three":["#101:2","#103:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}
0 | {"@type":"d","@rid":"#62:2","@version":3,"@class":"v2","in_one_two":[],"r":2,"out_two_three":["#102:2","#104:2"],"@fieldTypes":"in_one_two=g,out_two_three=g"}
0 | {"@type":"d","@rid":"#73:2","@version":1,"@class":"v3","r":1,"in_two_three":["#101:2","#104:2"],"@fieldTypes":"in_two_three=g"}
0 | {"@type":"d","@rid":"#74:2","@version":1,"@class":"v3","r":2,"in_two_three":["#102:2","#103:2"],"@fieldTypes":"in_two_three=g"}

[luigidellaquila]

Hi @tolgaulas

I just tried your script and I see a consistent graph at the end...

This is the expected behaviour, the update of out and in attributes of an edge also updates the linked vertices to maintain the graph consistent. Am I missing something?

Thanks

Luigi

[tolgaulas]

Hi @luigidellaquila , the screenshot is correct. The empty array in #58:0's in_one_two may cause problem with "is null" filter operations, doesn't it? If we set the new in with (select from #rid) notation the empty array does not show hence is null fillter operations looks like running fine.

[luigidellaquila]

Hi @tolgaulas

The empty array won't cause any problems, in general I'd suggest to use graph operators to check if edges exist, (eg. out().size() > 0) and avoid null checks on out_* and in_* attributes

Thanks

Luigi

[tolgaulas]

Thank you very much. Tolga Ulas +90 533 7464908, skypeid:tolgaulas

…

On Thu, Nov 28, 2019 at 2:59 PM luigidellaquila ***@***.***> wrote: Hi @tolgaulas <https://github.com/tolgaulas> The empty array won't cause any problems, in general I'd suggest to use graph operators to check if edges exist, (eg. out().size() > 0) and avoid null checks on out_* and in_* attributes Thanks Luigi — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#9071?email_source=notifications&email_token=AAK5MQS4UFCL3PBLTMH57L3QV6XCXA5CNFSM4JQAOLXKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFMMRTY#issuecomment-559466703>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAK5MQVVP4M6SOM74WELA3DQV6XCXANCNFSM4JQAOLXA> .