Issue 16213 - InvocationAdapterFactory doesn't check for outparametercount ->crash
Summary: InvocationAdapterFactory doesn't check for outparametercount ->crash
Status: CLOSED FIXED
Alias: None
Product: udk
Classification: Code
Component: code (show other issues)
Version: OOo 1.1 Beta2
Hardware: PC Linux, all
: P3 Trivial (vote)
Target Milestone: OOo 1.1 RC
Assignee: joergbudi
QA Contact: issues@udk
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2003-06-29 16:15 UTC by joergbudi
Modified: 2004-02-15 07:30 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---

Attachments
Suggestion to fix the crash (1.75 KB, patch)
2003-06-29 16:18 UTC, joergbudi 		no flags Details | Diff
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this issue.
Description joergbudi 2003-06-29 16:15:41 UTC 
Hi Daniel,

the invocation-adapter-factory does not check for the number of outparameters,
that are returned by the XInvocation.invoke() implementation. This leads to a
crash later in the bridge code, where the expected outparamters aren't initialized.

I yet can't provide sample code (too complicated), if you need it, tell me.

Joerg
Comment 1 joergbudi 2003-06-29 16:16:51 UTC 
It would be nice to get this fixed until OOo1.1rc, though it is not a
real problem for me.
Comment 2 joergbudi 2003-06-29 16:18:12 UTC 
Created attachment 7222 [details]
Suggestion to fix the crash
Comment 3 Daniel Boelzle [:dbo] 2003-06-30 10:03:24 UTC 
If this is not a real problem now, how about targetting 2.0?  I have
problems arguing for fixing this into oo1rc, because you wrote this is
not a real problem.  People tend to be more restrictive here nowadays
concerning last minute bug fixes (seems that they have learnt from the
past), which looks sensible to me for stable releases.  Last oo11rc
cws will be in one/two days.  I hope you agree or convince Kay :)
Comment 4 Daniel Boelzle [:dbo] 2003-06-30 14:34:04 UTC 
retargeted.
Comment 5 joergbudi 2003-07-01 19:41:13 UTC 
Hi,

actually targetting it to OOo2.0 is too late. KR, what's your opinion 
here ?

The bug hits you, if e.g. a python programmer implements an UNO 
object method with out parameters in the wrong way (e.g. an 
XInputStream, which is not so seldom). Once the programmer gets to 
know the reason for the crash, he can workaround it by fixing his 
wrong implementation. 

Neitherway, its a crash, and gives a potential third party developer 
a bad impression about office stability. If you need a longer test 
period for the fix, how about targeting it to 1.1 final or 1.1.x with 
x > 0 ? These targets don't yet exist, but will show up in future, I 
guess.

BTW, my suggested fix does not seem to be very risky IMHO ...

Bye, Joerg
Comment 6 kay.ramme 2003-07-02 17:36:13 UTC 
Joerg, you originaly said, that these is not a real problem for you.
So, just for the protocol, we are currently concentrating on real
problems :-). On the other hand, I tend to agree that this is a simple
and unrisky fix, but I think we don't want to have it for the final,
but may be 1.1.1 / PP1.
Comment 7 joergbudi 2003-07-03 06:27:48 UTC 
Ok, setting target to OOo 1.1.1, but we should remember to integrate
it  into the source tree as soon as it exists to have the longest
possible test phase.
Comment 8 kay.ramme 2003-07-07 12:01:09 UTC 
Daniel, as this is low risk and error handling only, please fix it in
1.1RC. We avoid having antoher library in the patch then.
Comment 9 Daniel Boelzle [:dbo] 2003-07-07 13:11:13 UTC 
@jbu: fixed in iafactory.cxx rev 1.10.18.1
      Please review/verify ASAP!
Comment 10 Daniel Boelzle [:dbo] 2003-07-07 13:11:55 UTC 
reassigned.
Comment 11 Daniel Boelzle [:dbo] 2003-07-07 13:20:29 UTC 
started.
Comment 12 Daniel Boelzle [:dbo] 2003-07-07 13:21:46 UTC 
resolved
Comment 13 joergbudi 2003-07-07 18:42:55 UTC 
Works fine for me. 

BTW, nice stacktraces in assertions feature :o)
Comment 14 Martin Hollmichel 2004-02-15 07:30:34 UTC 
close issue