MVEL

Variable expression calculate error

Details

  • Type: Bug Bug
  • Status: Closed Closed
  • Priority: Major Major
  • Resolution: Fixed
  • Affects Version/s: 1.3.5
  • Fix Version/s: 1.1
  • Component/s: Core
  • Labels:
    None
  • Environment:
    mvel 1.3.4
  • Number of attachments :
    4

Description

x = 20, y =10, z= 5

test

assertEquals(20-10-5, test("x-y-z")) failure

or

assertEquals(20/10/5, test("x/y/z")) failure

  1. Java Source File
    AbstractTest.java
    04/Jul/08 4:41 AM
    22 kB
    Ed.Wang
  2. Java Source File
    CoreConfidenceTests.java
    04/Jul/08 4:41 AM
    107 kB
    Ed.Wang
  4. File
    test_src.rar
    05/Jul/08 4:43 AM
    2 kB
    Ed.Wang

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Mike Brock added a comment -

Ouch. I'll get this fixed ASAP.

Show
Mike Brock added a comment - Ouch. I'll get this fixed ASAP.
Hide
Permalink
Mike Brock added a comment -

fixed in both trunks

Show
Mike Brock added a comment - fixed in both trunks
Hide
Permalink
Mike Brock added a comment -

closed.

Show
Mike Brock added a comment - closed.
Hide
Permalink
Ed.Wang added a comment -

The calculate result is not correct such as the next expression : a+b-c*x/y-z.

Show
Ed.Wang added a comment - The calculate result is not correct such as the next expression : a+b-c*x/y-z.
Hide
Permalink
Ed.Wang added a comment -

in my today download lib (ver:1.3.5)

Show
Ed.Wang added a comment - in my today download lib (ver:1.3.5)
Hide
Permalink
Ed.Wang added a comment -

Attach my test source code. The main test class is PerformanceTest.java. In this test generate random variable expression and then compare the mvel excute result with Rhino. Hope help.

Show
Ed.Wang added a comment - Attach my test source code. The main test class is PerformanceTest.java. In this test generate random variable expression and then compare the mvel excute result with Rhino. Hope help.
Hide
Permalink
Mike Brock added a comment -

fixed in release candidate. the release candidate is attached.

Show
Mike Brock added a comment - fixed in release candidate. the release candidate is attached.
Hide
Permalink
Ed.Wang added a comment -

still a lot of incorrect result, such as:

{b=487.0005392624622, c=758.7743539324256, a=986.4474715991837, z=178.55289227009197, y=356.4570150955031, x=166.7242753498912}

Error Occ1ur: x/b*a+a*z+b/b 882358.7934537028 880670.2473680319
Error Occ1ur: c/y*c 8075.847799366722 0.014026936736426366
Error Occ1ur: z/y/a*y/y/y*x 0.0011875377829837724 3.36229199811026E-13
Error Occ1ur: z/y*a+a 7402.844495287964 4932.239896955362
Error Occ1ur: y/a*a 1782.2850754775154 0.001831594127570113
Error Occ1ur: x/a*a/z+z 897.4332256933441 892.7644661483912
Error Occ1ur: c/y*c-a 3143.610441370804 -4932.223331059182
Error Occ1ur: y/z*b 4861.157844276353 0.02049655342235746
Error Occ1ur: c/y*a/y+a/z*a/x+x 1026.512926769278 997.0590928618801

please use my attachment to test it.

Show
Ed.Wang added a comment - still a lot of incorrect result, such as: {b=487.0005392624622, c=758.7743539324256, a=986.4474715991837, z=178.55289227009197, y=356.4570150955031, x=166.7242753498912} Error Occ1ur: x/b*a+a*z+b/b 882358.7934537028 880670.2473680319 Error Occ1ur: c/y*c 8075.847799366722 0.014026936736426366 Error Occ1ur: z/y/a*y/y/y*x 0.0011875377829837724 3.36229199811026E-13 Error Occ1ur: z/y*a+a 7402.844495287964 4932.239896955362 Error Occ1ur: y/a*a 1782.2850754775154 0.001831594127570113 Error Occ1ur: x/a*a/z+z 897.4332256933441 892.7644661483912 Error Occ1ur: c/y*c-a 3143.610441370804 -4932.223331059182 Error Occ1ur: y/z*b 4861.157844276353 0.02049655342235746 Error Occ1ur: c/y*a/y+a/z*a/x+x 1026.512926769278 997.0590928618801 please use my attachment to test it.
Hide
Permalink
Mike Brock added a comment -

fix applied to both trunks.

please be advised that I ran your test against the new code, and received the following errors:

—
Error Occur: b*z*x*c*z+a/a*z/y
Result1:4.32000006E8; Result2:4.32E8
—
Error Occur: y*c*b-a+z/y
Result1:149956.0; Result2149955.99609375
—
Error Occur: b*c*y+z/y/x/x*x
Result1:150000.15; Result2:150000.146484375

... but note that these differences are within tolerance. The results on the right represent the MVEL results, using rounding rules that comply with the IEEE 754R standard to a 128-bit precision (34 digits).

I have attached RC2

Show
Mike Brock added a comment - fix applied to both trunks. please be advised that I ran your test against the new code, and received the following errors: — Error Occur: b*z*x*c*z+a/a*z/y Result1:4.32000006E8; Result2:4.32E8 — Error Occur: y*c*b-a+z/y Result1:149956.0; Result2149955.99609375 — Error Occur: b*c*y+z/y/x/x*x Result1:150000.15; Result2:150000.146484375 ... but note that these differences are within tolerance. The results on the right represent the MVEL results, using rounding rules that comply with the IEEE 754R standard to a 128-bit precision (34 digits). I have attached RC2
Hide
Permalink
Mike Brock added a comment -

attached 1.3.6RC2

Show
Mike Brock added a comment - attached 1.3.6RC2
Hide
Permalink
Ed.Wang added a comment -

Mike, thanks you great work.

Show
Ed.Wang added a comment - Mike, thanks you great work.
Hide
Permalink
Mike Brock added a comment -

closed

Show
Mike Brock added a comment - closed

People

  • Assignee:
    Mike Brock
    Reporter:
    Ed.Wang
Vote (0)
Watch (0)

Dates

  • Created:
    Updated:
    Resolved: