[SERVER-5108] find7.js is failing in V8 build Linux 64 Created: 27/Feb/12  Updated: 11/Jul/16  Resolved: 29/Feb/12

Status: Closed
Project: Core Server
Component/s: Testing Infrastructure
Affects Version/s: None
Fix Version/s: 2.1.1

Type: Bug Priority: Major - P3
Reporter: Eric Milkie Assignee: Michael A. Fiedler
Resolution: Done Votes: 0
Labels: buildbot
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
is related to SERVER-5034 strange object behavior in v8 shell Closed
Operating System: Linux
Participants:

 Description   

Mon Feb 27 11:47:47 [conn355] build index done.  scanned 0 total records. 0 secs
assert: [{ "_id" : { "d" : 3649, "w" : "signed" }, "u" : { "3649" : 5 } }] != [{
	"_id" : {
		"d" : 3649,
		"w" : "signed"
	},
	"u" : {
		"3649" : 5,
		"3649" : 5
	}
}] are not equal : A1
Error: Printing Stack Trace
    at src/mongo/shell/utils.js:37:7
    at src/mongo/shell/utils.js:58:1
    at Function.eq (src/mongo/shell/utils.js:84:1)
    at /data/slave/Linux_64bit_v8/mongo/jstests/find7.js:6:8



 Comments   
Comment by Michael A. Fiedler [ 29/Feb/12 ]

This is now running, and has passed find7.js.

Antoine pointed out that he was using a particular revision - r10841 - which corresponds to an in-development version of v8's trunk, 3.9.11.
v8 has a release branch for prior versions, such as 3.8, etc - and 3.9.x is simply not "stable", at least not between tagged versions. The tag is not a real tag, simply a commit with a "tagged" message.

I would suggest that we figure out which set of things we want from v8, and determine a branch/update model to prevent this underlying dependency from breaking and consuming this much time again.

Comment by Tad Marshall [ 28/Feb/12 ]

Cool. I think it's time to put V8 in src/third_party and stop letting differences in machine setups dictate which bugs show up on which machines. The Jira ticket to do that should be separate from this one; I can issue it if you want.

Comment by Antoine Girbal [ 28/Feb/12 ]

Ok I do see the bug with that version of V8, I think it's their issue 1772.
Could you run on builbot v8 folder:

$ svn up
$ scons -j3 arch=x64 debuggersupport=off snapshot=off profilingsupport=off

should pass afterwards

Comment by Michael A. Fiedler [ 27/Feb/12 ]

Here's the goods:

URL: http://v8.googlecode.com/svn/trunk
Repository Root: http://v8.googlecode.com/svn
Repository UUID: ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Revision: 10579
Node Kind: directory
Schedule: normal
Last Changed Author: danno@chromium.org
Last Changed Rev: 10576
Last Changed Date: 2012-02-01 10:48:36 +0000 (Wed, 01 Feb 2012)

Comment by Tad Marshall [ 27/Feb/12 ]

I think he wants the equivalent of the Git hash, but for the Subversion repo that the V8 code is in.

> svn info
Path: .
Working Copy Root Path: C:\Users\Tad\Documents\Visual Studio 2010\Projects\v8
URL: http://v8.googlecode.com/svn/trunk
Repository Root: http://v8.googlecode.com/svn
Repository UUID: ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Revision: 10816
Node Kind: directory
Schedule: normal
Last Changed Author: kmillikin@chromium.org
Last Changed Rev: 10802
Last Changed Date: 2012-02-23 03:45:21 -0500 (Thu, 23 Feb 2012)

Comment by Eric Milkie [ 27/Feb/12 ]

I can't login but Michael can

Comment by Antoine Girbal [ 27/Feb/12 ]

ok that's odd, I saw that problem but then fixed it and it doesnt happen for me on my box.
Could you give me the v8 version on this bbot (svn info in V8 folder)?

Generated at Thu Feb 08 03:07:55 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.