62282 - API tests fail for NULL values of parameters for specific methods.

Status: RESOLVED INCOMPLETE

(Core Graveyard :: Java: OJI, defect, P3)

Description

[rpallath]

Below is  a list of methods that fail for NULL values of parameters.
They can be executed using OJI API testswhich lie in dir. mozilla/modules/oji/tests.
(Look at README out there for build/execution).
(If referring to internal SUN page visit
http://lonely.eng.sun.com/Blackwood/doc/OJI/index.html for getting the dynamic
libraries to test these API's)

These test cases can not be simulated by end-user thru' a normal APPLET operation.
The API tests are designed to test each method for all valid and invalid
conditions and use XPconnect mechanism thru' javaScript to invoke the methods.

Fix could be as simple as just checking for NULL conditions.
List of methods that fail

NIEnv_GetBooleanField_1
JNIEnv_GetByteField_1
JNIEnv_GetCharField_1
JNIEnv_GetDoubleField_1
JNIEnv_GetFloatField_1
JNIEnv_GetIntField_1
JNIEnv_GetLongField_1
JNIEnv_GetMethodID_5
JNIEnv_GetObjectField_1
JNIEnv_GetShortField_1
JNIEnv_GetStaticBooleanField_1
JNIEnv_GetStaticByteField_1
JNIEnv_GetStaticCharField_1
JNIEnv_GetStaticDoubleField_1
JNIEnv_GetStaticFloatField_1
JNIEnv_GetStaticIntField_1
JNIEnv_GetStaticLongField_1
JNIEnv_GetStaticObjectField_1
JNIEnv_GetStaticShortField_1
JNIEnv_NewStringUTF_3
JNIEnv_SetBooleanField_1
JNIEnv_SetByteField_1
JNIEnv_SetCharField_1
JNIEnv_SetDoubleField_1
JNIEnv_SetFloatField_1
JNIEnv_SetIntField_1
JNIEnv_SetLongField_1
JNIEnv_SetShortField_1
JNIEnv_SetStaticBooleanField_1
JNIEnv_SetStaticByteField_1
JNIEnv_SetStaticCharField_1
JNIEnv_SetStaticDoubleField_1
JNIEnv_SetStaticFloatField_1
JNIEnv_SetStaticIntField_1
JNIEnv_SetStaticLongField_1
JNIEnv_SetStaticObjectField_1
JNIEnv_SetStaticShortField_1
JNIEnv_ThrowNew_4


(see attachmt for description of those tests).

Tested using FCS bits of Netscape 6(commercial) dated 12/04/2000 on Sparc
2.7/2.8 and intel 2.7/2.8
using JRE 1.3.0_01  (FCS version)

Comment 1

[rpallath]

Attachment: HTML file describing the tests.

Comment 2

[edburns]

Joe, can you please look at this one?

Comment 3

[rpallath]

Updated QA contact and added avm to CC list

Comment 4

[Alexei V. Mokeev]

These tests are failed under Linux too (with 2001040805 build)

Comment 5

[Alexei V. Mokeev]

Also all these tests are PASSED under WinNT SP6, build from 04/14/2001.
jre1.3.0_01 was used under WinNT and Linux

Comment 6

[Alexei V. Mokeev]

All these tests are FAILED for me under Solaris 8
jre 1.3.1 and recent mozilla trunk.
Attached please see .tar.gz file with logs.

Comment 7

[Alexei V. Mokeev]

Attachment: Logs in tar.gz format

Comment 8

[Alexei V. Mokeev]

Behavior with jre1.4.0-beta similar. 

Comment 9

[shrirang khanzode]

SPAM: reassigning all OJI bugs to new OJI QA, pmac ( 227 bugs)

Comment 10

[Raghu Sakleshpur]

Alexei,

     I'm assigning you as the owner.

-raghu

Comment 11

[Igor Nekrestyanov]

I guess this is not OJI issue. 

These routines provide JNI functionality and therefore should 
conform to JNI spec
(see http://java.sun.com/j2se/1.3/docs/guide/jni/spec/functions.doc.html)

In particular spec says 

"Note the use of the term "must" to describe restrictions on JNI programmers.
For example, when you see that a certain JNI function must receive a non-NULL
object, it is your responsibility to ensure that NULL is not passed to that JNI
function. As a result, a JNI implementation does not need to perform NULL
pointer checks in that JNI function. "
 
And 
 for Get<type>Field Family and Set<type>Field Family Routines 
 spec explicitly says 
  "obj: a Java object (must not be NULL)."

Therefore i think at least for these methods bug is invalid.

For rest of methods i did not find any details on expected behavior
in the spec. 
2Reporter: which spec you used to specify expected behaviorfor these methods? 

Anyway, all these funcs forward requests to corresponding JNI methods and 
therefore this is problem of used JNI implementation.

Comment 12

[Benjamin Smedberg]

Mass-closing bugs in the "OJI" component: OJI plugin integration was replaced with npruntime long ago, and these bugs appear to be irrelevant now. If there is in fact a real bug that remains, please file it new in the "Core" product, component "Plug-ins".