Skip to main content

Non-regression tests

The Tests folder checks that JD Edwards still behaves the same after a change — an ESU, a tools release, an upgrade, a package build. A test calls JD Edwards the way an integration does, through the AIS server: it opens a form, reads a table, runs a business function, and compares the answer with what is expected and with a reference run recorded before the change.

Nothing is installed in JD Edwards, and no script is written: a case is built by picking calls in a list and letting JD Edwards' own answer fill in what to expect.


At a glance​

TestsSaveRun this caseRun the suiteRecord the referencealso the cases that write+ New suitesmokeais-is-upudc-types-of-system-00address-book-formunknown-table-is-refused+ New caseaddress-book-formDATA_QUERYread≥ 1 row(s) · SY = 00+FORM_P01012_W01012Bdescribeworks · 12 value(s) keptEndpassedchanged: differsfrom the referenceTHE STEPENDPOINT CALLEDdata_query ▾PARAMETERSTABLE F0004WHAT IS EXPECTEDFill from the answerThe callmust workRows, at least1The first row holdsSY = 00Suites and caseswith their last resultA case is a flowone card per callThe + adds a stepat that placeRecord the referencebefore the changeFill from the answerproposes what to expect

The three entries​

Menu entryUse
Tests → ResultsHow the last runs went: cases passed, failed, changed or never played, the last runs of each suite as bars, each case with its last results.
Tests → CasesThe suites and their cases, drawn and edited as flows of calls.
Tests → Run the testsThe job that plays a suite — on demand here, or on a schedule.

Results and Cases are the test pages of Liberty Next, described in full in Tests; this page covers what is specific to JD Edwards.


A case, step by step​

A suite holds cases; a case is a sequence of steps; a step is one call to JD Edwards with what is expected of its answer.

  1. In Tests → Cases, pick the suite and choose New case. Say what it checks, and tick This case changes data when one of its steps writes in JD Edwards.
  2. The + on the flow adds a step. Pick the Endpoint called in the list of the AIS connector: one of the catalogue's services — a table read, a business function — or an endpoint made with Create from… for a form, a report or an orchestration.
  3. Fill the Parameters: one labelled field per parameter of the endpoint, with its help.
  4. Fill from the answer plays the call now and proposes what to expect — the number of rows and the values of the first row, by JD Edwards alias (AN8, ALPH). Keep the lines that must stay true, delete the others.
  5. Save, then Run this case.

What a step can expect:

ExpectationMeaning
The call must work / must be refusedA refusal is a valid expectation: a table that does not exist, a value the form rejects. The refusal says checks the message.
Rows, at least / exactly / at mostThe number of rows returned.
The first row holdsValues of the first row, by column name.
A warning saysA warning of JD Edwards (Duplicate Address Book record) reported with the answer.

A value read at one step — an address number just created — can be kept and given to the next steps. A step marked Played even after a failure (a cleanup) always runs: this is how a case that writes removes what it created.


Reference and comparison​

Expectations say what must be true. The reference catches what nobody thought of writing down: the columns of a form, the labels of a grid, the values of a set-up table.

  1. Before the change, record the reference: Record the reference on the Cases page, or the job with record. Each step keeps the parts of its answer it was asked to keep.
  2. After the change, run the suite again. Each case ends as:
ResultMeaning
PassedEvery expectation holds and nothing differs from the reference.
ChangedThe expectations hold, but a kept value differs from the reference — a column gone, a new one, another value. The difference is listed value by value.
FailedAn expectation is not met, or the call failed.
SkippedThe case writes on JD Edwards and the run did not allow it.

A Changed is not always a defect — an update may legitimately add a column — but it is always something to read. Record the reference again once the change is accepted.


The job​

Tests → Run the tests plays a suite outside the page — on demand, or on a schedule after each package deployment.

ParameterDefaultMeaning
SuitesmokeThe suite to play, among the suites of the AIS connector.
Compare with the reference, or record itcomparecompare checks the cases and reports what differs; record keeps the run as the new reference.
Only these casesblankCase identifiers separated by commas; blank plays them all.
Cases that write on JDEnoyes also plays the cases marked as changing data.

The run fails when a case fails or has changed, which triggers the job's alerts. Its files hold an Excel report, one line per step: case, label, case status, step, endpoint, step status, duration and what was found. The results also appear in Tests → Results.


The suite shipped​

Nomajde ships a first suite, smoke — JDE — what must keep working — made of reading cases any JD Edwards answers:

CaseChecks
ais-is-upAIS answers and says what it runs.
udc-types-of-system-00The UDC types of system 00 are read from F0004.
address-book-formThe Address Book search form opens and still has its columns.
unknown-table-is-refusedA table that does not exist is refused, not answered empty.

Use it to check the connection, then add the cases that matter to the business: the forms the users live in, the business functions the interfaces call, the set-up tables an update must not touch.


Access​

ToA role needs
See the suites and the resultsSee the test suites and their results
Play a case or a suitePlay the tests (and see them)
Write suites, cases and stepsWrite the tests: suites, cases, steps (and play them)
Run the job from the menuThe job's permission

The three test permissions are in a role's Beyond the menus rules, type Tests: a functional team can write and play the tests without being administrators. Every call runs under the JD Edwards account of the AIS connector: JD Edwards security applies to that account.


Tips & best practices​

  • Record the reference on the environment that will be changed, just before the change.
  • Prefer many small reading cases to a few long ones: a failure then names what broke.
  • Keep the writing cases apart, each with its cleanup step, and play them on a test environment.
  • Schedule the job after the nightly package deployment: a regression is known in the morning, with the difference listed.
  • The AI assistant knows the procedure: ask it to check that JDE still behaves the same after a change.