Selected code for subcomponent as other choices seemed inadequate.
Feel free to change this.

I have read the FAQ. I know how to do java. The installation program
is broken.

I can't reproduce: Selecting 'Do not use....' brings up the message
box 'Some functionality....'. Klicking on >ok< in this messagebox and
the >ok< button in java setup is enabled and setup is running fine.

Not reproducible.

Wonderful! We know that it OUGHT to work. However, there's soemthing
in the code that causes it to fail under M$W98SE. As I explained, I
first tried a full install (1U) and when it failed to find Java, I
stopped the process and tried again from scratch. One the second
attempt, I tried the "Recommended" install which, I think, leaves out
xlst and friends so no java should be needed. However install again
found no java install and upon selecting so I clicked "ok" proceeded
to select the "no java" button and install promptly blanked the the
"no java" radio button and demanded the java be installed. This is not
the behaviour of a well constructed program. So if java is a
requirement under the circumstances I have described then selecting
"no java" should reset things so that install can proceed. I perceive
that the process you used assumes prior knowledge that is not well
documented and is certainly not intuitive.

I am re-opening the bug.

Improper methodology is no reason to state unconfirmed. If this occurs
on release, we got a problem.

So what's the problem here? OOo setup starts with 'used an
installed...'. That's ok, if java has been detected it will be
displayed, if not, the user has the opportunity to browse to a usable
java location. If there is no java, the user can set the radio button,
he get's a warning and can go on.

I don't understand. You insist that it offers a chance to go on. IT
DOES NOT! The install loops between looking for java, saying it can
find it via a popup, you click okay, you select install without, a
popup comes up saying you lose functionality, you click ok, install
blanks the no java radio button and refuses to go on.

If oo.org did what you claim then there'd be no problem. That is does
not do what you claim is the problem. What's so hard to understand? We
are talking apples and oranges and must meet on common ground.

No matter for screaming around here.
On my Win98 machine it behaves as descibed by me. Java setup starts
'use an installed...' and if I choose 'Don't use java...' setup goes
on, the >ok< button in java setup is enabled.

Windows 98 is not windows 98 SE, agreed. I thought I specified this.
It was a raw install so no java. Please try it. Perhaps in your case
there is register variable that permits install to work. In my it did
not. Anyways I have neither the time nor energy to pursue this issue.
Please do what you want. I would hate to see this appear in subsequent
releases.

I have taken the time and I've checked on Win98 SE, WinME, Win2000, XP
and Linux and I can't reproduce. What else shall I do?

Whatever, I have changed the status. If it bites later ...

No need to go on. I accept that this will not be fixed.

I had no problems with linux and RC3, but I wasn't trying a laptop
(what kind).  Did you do a setup.exe -net before running a straight
setup.exe?  I am just guessing it would be the cause of the problem. 
But I am a mere user, not a developer.  Because OOo has installed
without a hitch for so long, I assume it has something to do with the
fact that you used a laptop.

Did the developer test with a laptop and Windows 98?  Perhaps the
combination has something to do with it.

This is the same 19318. Reopening.

Reopening for real this time

Please close 19318 as a duplicate and update the status.

*** Issue 19318 has been marked as a duplicate of this issue. ***

I've checked again with RC4 on Win98SE and still can't reproduce. Is
there any chance to see in the code why this might happen?

I can only describe the platform on which this occurs. I do so because
I think there may be a relationship to the hardware.

My daughter's laptop is an old Acer (TI) with 64Mb of memory. It's not
at hand but I believe it's a Pentium2 @ 166 MHz. Lot's of disc (3 Gb)
running windows 98 SE with no updates. RC3 image md5sum matches the on
the download page.

This is the best I can do from memory; however the looping is
repeatable. Please ask for more info on this and the other system. 

Is the problem in the sources, probably, but it's tickled by the
target configuration is the nearest I can get at this point.

There is an error in the focus handling of message boxes called from
modal dialogs.

To grsingleton:

could you please check if the problem disappears when you use the
keyboard to select the 'don't use java' button?

Thanks, Dirk

The following code snippet in winproc.cxx will fix this 'modal dialog'
bug. It should not be possible to activate a disabled dialog. I'll
check this in as soon as I have a PP1 workspace.

           BOOL bHandled = FALSE;
           if ( mpFrameData->mpFocusWin->IsInputEnabled() )
           {
               if ( mpFrameData->mpFocusWin->IsEnabled() )
               {
                   mpFrameData->mpFocusWin->GrabFocus();
                   bHandled = TRUE;
               }
               else if( mpFrameData->mpFocusWin->ImplHasDlgCtrl() )
               {
               // #109094# if the focus is restored to a disabled
dialog control (was disabled meanwhile)
               // try to move it to the next control
                   mpFrameData->mpFocusWin->ImplDlgCtrlNextWindow();
                   bHandled = TRUE;
               }
           }
           if ( !bHandled )
           {
               ImplSVData* pSVData = ImplGetSVData();
               Window*     pTopLevelWindow =
mpFrameData->mpFocusWin->ImplGetFirstOverlapWindow();
               if ( !pTopLevelWindow->IsInputEnabled() &&
pSVData->maWinData.mpLastExecuteDlg )
                   pSVData->maWinData.mpLastExecuteDlg->ToTop(
TOTOP_RESTOREWHENMIN | TOTOP_GRABFOCUSONLY);
               else
                   pTopLevelWindow->GrabFocus();
           }

I will be able to check tonight when I meet with my daughter. I have
only rc3 with which to check. Is this a problem?
http://distribution.openoffice.org/cdrom/

To grsingleton:

Testing with RC 3 is ok. For me it looks like you somehow produced two
mouse click events. At least this has been the only way for me to
reproduce your bug. So if selecting 'no java' with the keyboard works,
we have found ( and fixed ) your bug.

Thanks, Dirk

I looked at the code snippet and would agree. I was not able to get
the laptop yesterday but will test asap. Is the weekend too late?

I have not been able to get my hands on the offending laptop. Please
bear with me.

Fixed in CWS vcl7pp1r3.

Just a quick note to say that the workaround worked. I look forward to
1.1.1.

BTW, I have pointed a linux user at this issue and the workaround
worked for him also. Here's a copy of his last message:

On Wed, 05 Nov 2003 09:52:08 -0500, G. Roderick Singleton
<grsingleton@openoffice.org> wrote:

>>>>>> I click the button again and get the same error.  There is no
way out
>>>>>> other than to cancel the installation.
>>>>>>
>>>>>> I tried to install Java and it just locks up my machine just
after it 
>>>>>> says it is doing a checksum.
>>>>>>
>>>>>> Is it possible to install OpenOffice on Linux without Java being 
>>>>>> installed?
>
>> 
>> Maybe maybe not. http://www.openoffice.org/issues/show_bug.cgi?id=18584
>>

Going through the installation not using my mouse worked.  What needs
to be done as a followup?

This system is as follows:

Operating System: Slackware 9.0
Motherboard: Epox 8K7A+
Processor: AMD Athlon
Memory: 512MB
Window System: FVWM2
glibc Version: 2.3.1
Mouse: Logitech Optical Mouse (Has a USB connector on the end of the
cord.  It came with a USB-PS2 adapter and that is what I am using.) 

I can get additional information if needed.

I also tried another download site and had the same problem as before.

Is it possible the mouse is the problem?  I have had several lockups
with both Linux and Windows 2000 when clicking on web pages, etc.
These are total lockups and nothing can get the machine back except to
hit the Reset switch.

I am guessing that you are the grsingleton listed in the messages.  It
says that it was "Fixed in CWS vcl7pp1r3."  What does this mean?  I am
guessing that this was after the release date of 1.1.0?

Thanks for the help.
-- Frank Hahn God is not dead! He's alive and autographing bibles at
Cody's 

Hello:

I wanted to reconfirm Mr. Singletons response below that I could not
override the Java requirement on installation when using a mouse for
input.  When I used my keyboard, I was allowed to make the selection
that I did not have Java installed.

I have an Optical Logitech mouse which may or may not be a problem. 
This was also on a Slackware Linux installation.

I wanted to changed a couple of the items above, for example:

OS: from Windows 98 to Linux (Would not let me change)
Version: OOo 1.1 RC3 to OOo 1.1

Thanks.

Frank Hahn

ssa->of: please verify.

Seems IZ has ISSUES. I attempted to update this issue a number of days
a WRT my experience with the laptop but obviously it was posted.

Here goes again:

Using the keyboard on the laptop did allow me to get by the problem as
you predicted. However I would note that you must NOT use the mouse at
all until you get past the Java part. After is okay.

New focus behaviour at java installation.

Verified in cws vcl7pp1r3.

Ok in srx645m27.