0033133: OBCriteria skip adding client+org filters if table is known to not required these.

Openbravo Issue Tracking System - Openbravo ERP

View Issue Details

Project

Category

View Status

Date Submitted

Last Update

0033133

Openbravo ERP

A. Platform

public

2016-06-03 17:14

2022-02-01 08:09

Reporter

shuehner

Assigned To

Triage Platform Base

Priority

normal

Severity

minor

Reproducibility

have not tried

Status

new

Resolution

open

Platform

OS Version

Product Version

Target Version

Fixed in Version

Merge Request Status

Review Assigned To

OBNetwork customer

Web browser

Modules

Core

Support ticket

Regression level

Regression date

Regression introduced in release

Regression introduced by commit

Triggers an Emergency Pack

Summary

0033133: OBCriteria skip adding client+org filters if table is known to not required these.

Description

Every table in Openbravo is configured with a Data Access Level (in ad_table).

For tables which are 'System only' adding filters for readable clients + organizations should not be required.

So depending on the table used for a criteria those filters should be skipped to avoid some processing overhead.

Any change here needs to be carefully validated to not break client+org security model.
Also there maybe special cases (table joins) which probably require extra care.

NOTE: Even some partial optimization for some cases (i.e. no joins) could be useful.

Steps To Reproduce

Measure total runtime of i.e. OBMessageUtils.messageBD unmodified versus a version of the function which 'manually' removes those extra filters.

To simulate change the following can be used:
      obcMessage.setFilterOnReadableClients(false);
      obcMessage.setFilterOnReadableOrganization(false);

As a simple testcase to show performance difference the following can be used to do 10 runs of 10k invocations of the function with and without the proposed change.

  @Test
  public void testSHU2() {
    for (int run = 0; run < 10; run++) {
      long s1 = System.currentTimeMillis();
      for (int i = 0; i < 10000; i++) {
        String res = OBMessageUtils.messageBD("OrderDocumentno");
      }
      long s2 = System.currentTimeMillis();
      log.info("Took: " + (s2 - s1));
    }
  }

Testing in a normal pristine pi + retail database show a runtime of
a.) unmodified: ca. 8.4 seconds per run
b.) patched: 6.8s per run

In a different systems having a lot more organizations (ca. 300) the following time difference can be seen:
a.) unpatched: ca. 12.1 seconds per run
b.) patched: ca. 10.2s per run

Proposed Solution

Additional Information