View Issue Details
ID | Project | Category | View Status | Date Submitted | Last Update |
---|---|---|---|---|---|
0027710 | Open CASCADE | OCCT:Tests | public | 2016-07-25 17:28 | 2019-07-13 14:31 |
Reporter | Assigned To | bugmaster | |||
Priority | normal | Severity | minor | ||
Status | closed | Resolution | fixed | ||
Target Version | 7.4.0 | Fixed in Version | 7.4.0 | ||
Summary | 0027710: Make Data-exchange test cases more machine independent | ||||
Description | In the current version of DE-test scripts (including product test cases such as Parasolid, SAT...), comparison with reference data (especially floating points values, such as TOLERANCE) is made with constant tolerance (e.g. 1.0e-7). However, complex OCCT-algorithms often cannot provide this precision and result is often different on LINUX and WINDOWS platforms. At that, delta can be greater than 1.0e-7. More over, if current shape tolerance (value got from the file) is ~ 1.0e-4 then the fluctuations in range 2...3e-7 seems to be OK. My offer is following: 1. Enter new variable (e.g. MaxFloatingError). 2. Define this variable in all "begin" files with default value. 3. Redefine this variable in several test scripts (where cross-platform differences are detected) directly. 4. Use this variable (instead of constant value 1.0e-7) in all "end" files. P.S. This solution is considered to be implemented in both OCCT DE-test scripts and OCCT-products. It allows reducing number of variables such as "LinuxDiff" and "LinuxFaulties" (see de\step_4\B3 test case). At that, this-variables-mechanism is used in OCCT only (it is absence in OCCT-products). | ||||
Steps To Reproduce | N/A | ||||
Tags | No tags attached. | ||||
Test case number | Not required | ||||
|
Branch CR27710 has been created by apn. SHA-1: 70760da0ad61f8467fc235de7d968a333252ab07 Detailed log of new commits: Author: apn Date: Mon Sep 5 17:44:05 2016 +0300 0027710: Make Data-exchange test cases more machine independent Enter variable MaxFloatingTol to use it instead of constant value in end file. |
|
Dear nbv, could you please review OCCT branch CR27710 and PRODUCTS branch CR27710_1. |
|
Dear Sergei, Please check, if all necessary grids are changed in CR27710_1 branch (for OCCT-products). Are there any lost grids? TIA. |
|
I see that the implementation of the fix is incomplete, because the point 3 of the proposed steps has not been done: 3. Redefine this variable in several test scripts (where cross-platform differences are detected) directly. It is needed to identify all non-stable test cases and set necessary value of MaxFloatingError there. |
|
Dear Colleagues, This feature is entered in order to avoid regressions connected with numerical instability by the reason of different floating-point-number representations on Windows and Linux platforms. I.e. it does not mean, that we must insert new variable in all test cases unthinkingly (in automatic mode). The reason of the difference between numbers must be investigated individually in every test case. Especially, if these differences will be very big we shall understand the reason more carefully (enter the tolerance in this test case is wrong solution). Dear Andrei, According to above said, please make following: 1. Check all TODOs in unstable test cases. If this TODO refers to the real bug and this bug describes real reason of differences then we shall fix it (not to hide by the tolerance). 2. Please write something about made changes in User Guides and in commit messages. Otherwise, nobody will know about new feature. |
|
Branch CR27710 has been updated forcibly by apn. SHA-1: d03be33f0d8f5926dd8e3bdd8584c2f44da3b79c |
|
Branch CR27710 has been updated by apn. SHA-1: a807bfecbe5a7aa39d83122f1ad796ad6ce1dcbe Detailed log of new commits: Author: apn Date: Thu Sep 8 15:20:43 2016 +0300 Added description of MaxFloatingTol in user guides. |
|
Dear nbv, There are no unstable test cases is de group in occt. In Products: cr standard E1 - unstable test case. Commit message was corrected. Description of MaxFloatingTol was added in user guides. |
|
Dear Artem, Please check, if all necessary grids are changed in CR27710_1 branch (for OCCT-products). Are there any lost grids (see message 0027710:0057479)? TIA. |
|
Canonical Recognition is not part of Data Exchange. Thus it's tests doesn't need to be updated. All other grids are fine. |
|
Dear anv, remarks were applied (changes in cr were reverted). |
|
I wonder what was the reason of this bug creation, if eventually no one test case was updated. |
|
Branch CR27710_1 has been created by abv. SHA-1: f283875beef39a582f023ccc6f3de25d09cc4851 Detailed log of new commits: Author: apn Date: Thu Sep 8 15:20:43 2016 +0300 0027710: Make Data-exchange test cases more machine independent Added description of de test grid in user guide. |
|
Branch CR27710 has been deleted by abv. SHA-1: a807bfecbe5a7aa39d83122f1ad796ad6ce1dcbe |
|
I propose integrating branch CR27710_1 containing only change in documentation (added description of test group "de"). Other changes made within this issue do not seem to have any effect. The problems in tests should be addressed in more specific way. |
|
OK. |
|
Branch CR27710_1 has been deleted by inv. SHA-1: f283875beef39a582f023ccc6f3de25d09cc4851 |
Date Modified | Username | Field | Change |
---|---|---|---|
2016-07-25 17:28 |
|
New Issue | |
2016-07-25 17:28 |
|
Assigned To | => apn |
2016-07-25 17:29 |
|
Description Updated | |
2016-09-05 17:42 | git | Note Added: 0057472 | |
2016-09-05 17:57 | apn | Note Added: 0057473 | |
2016-09-05 18:21 | apn | Assigned To | apn => nbv |
2016-09-05 18:21 | apn | Status | new => resolved |
2016-09-05 18:35 |
|
Note Added: 0057479 | |
2016-09-05 18:35 |
|
Assigned To | nbv => szv |
2016-09-05 18:35 |
|
Status | resolved => feedback |
2016-09-06 09:53 |
|
Note Added: 0057491 | |
2016-09-06 09:53 |
|
Assigned To | szv => apn |
2016-09-06 09:53 |
|
Status | feedback => assigned |
2016-09-06 09:58 |
|
Assigned To | apn => nbv |
2016-09-06 09:58 |
|
Status | assigned => feedback |
2016-09-06 10:49 |
|
Note Added: 0057493 | |
2016-09-06 10:49 |
|
Assigned To | nbv => apn |
2016-09-06 10:49 |
|
Status | feedback => assigned |
2016-09-08 12:24 | git | Note Added: 0057591 | |
2016-09-08 15:18 | git | Note Added: 0057594 | |
2016-09-08 15:24 | apn | Note Added: 0057595 | |
2016-09-08 15:25 | apn | Assigned To | apn => nbv |
2016-09-08 15:25 | apn | Status | assigned => feedback |
2016-09-08 15:31 |
|
Note Added: 0057596 | |
2016-09-08 15:34 |
|
Assigned To | nbv => anv |
2016-09-08 15:47 | anv | Assigned To | anv => apn |
2016-09-08 15:49 | anv | Note Added: 0057598 | |
2016-09-08 15:49 | anv | Status | feedback => assigned |
2016-09-08 16:29 | apn | Note Added: 0057601 | |
2016-09-08 16:30 | apn | Assigned To | apn => anv |
2016-09-08 16:30 | apn | Status | assigned => resolved |
2016-09-08 16:37 |
|
Assigned To | anv => nbv |
2016-09-08 17:12 |
|
Note Added: 0057605 | |
2016-09-08 17:59 |
|
Assigned To | nbv => apn |
2016-09-08 17:59 |
|
Status | resolved => assigned |
2016-11-07 14:08 | apn | Target Version | 7.1.0 => 7.2.0 |
2017-07-24 10:34 | bugmaster | Target Version | 7.2.0 => 7.3.0 |
2018-02-25 21:09 |
|
Target Version | 7.3.0 => 7.4.0 |
2019-07-11 09:28 | git | Note Added: 0085582 | |
2019-07-11 09:30 | git | Note Added: 0085584 | |
2019-07-11 09:32 |
|
Note Added: 0085585 | |
2019-07-11 09:32 |
|
Assigned To | apn => msv |
2019-07-11 09:32 |
|
Status | assigned => resolved |
2019-07-11 09:32 |
|
Steps to Reproduce Updated | |
2019-07-11 10:31 |
|
Note Added: 0085589 | |
2019-07-11 10:31 |
|
Assigned To | msv => bugmaster |
2019-07-11 10:31 |
|
Status | resolved => reviewed |
2019-07-11 11:00 | bugmaster | Test case number | => Not required |
2019-07-11 11:06 | bugmaster | Status | reviewed => tested |
2019-07-13 14:17 | bugmaster | Changeset attached | => occt master 943b4e5b |
2019-07-13 14:17 | bugmaster | Status | tested => verified |
2019-07-13 14:17 | bugmaster | Resolution | open => fixed |
2019-07-13 14:31 | git | Note Added: 0085618 |