Issue 17160 - Installer crashes consistently in the final phase of installation
Summary: Installer crashes consistently in the final phase of installation
Status: CLOSED DUPLICATE of issue 15288
Alias: None
Product: Installation
Classification: Application
Component: code (show other issues)
Version: OOo 1.1 RC
Hardware: PC Windows 98
: P1 (highest) Trivial (vote)
Target Milestone: ---
Assignee: Olaf Felka
QA Contact: issues@installation
URL: Not applicable
Keywords:
Depends on:
Blocks:
 
Reported: 2003-07-21 05:15 UTC by wg3
Modified: 2003-09-08 16:53 UTC (History)
1 user (show)

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


Attachments

Note You need to log in before you can comment on or make changes to this issue.
Description wg3 2003-07-21 05:15:00 UTC
I have tried to run the installer four times as follows:
  1. Fresh on new OS installation. Results type A1.
  2. After rebooting and running Scandisk, tried to restart. Results type B1.
At this point, I exported the Windows registry and manually deleted all 
references to the program therein, AND manually deleted the program subdirectory 
as well.
  3. Fresh after actions just described. Results type A2.
  4. After rebooting and running Scandisk, tried to restart. Results type B2.

After trying #4, I noticed that, despite the failure of the install program to 
finish, the program in fact appears in the Start Menu and starts enough to 
present a work screen. That is as far as I have taken this issue thus far.

Results code A:
  There are five phases involved in the installation procedure. The program 
consistently got as far as the third time through the progress indicator in 
Phase 5, which is called something like "Creating local settings" (pardon me, 
but my memory is not 100% reliable, and this is frustratingly hard to reproduce 
-- perhaps it is better to call it impossible in my circumstances), and both 
times the installer locked up at precisely the same point in the processing. I 
had to CTRL-ALT-DEL the machine to regain control. Each time Windows said the 
program was "not responding." Termination appeared successful, but in fact was 
not. Each time a failure-reporting program appeared and offered to send a 
failure report in for me; each time this operation failed.

Variation #1: I tried to "view the report", and that failed. CTRL-ALT-DEL also 
failed after an initial appearance of success: the screen was not refreshed, and 
indeed the machine had locked up so badly that the reset button was needed to 
regain control.

Variation #2: This time, trying to "view the report" was successful. However, 
trying to send it was not. Obviously the report was huge. Again the 
failure-reporting program had to be canceled with CTRL-ALT-DEL; this time it was 
successful in that I was able to regain control of the machine without pressing 
the reset button. I immediately rebooted. On a hunch, I ran SCANDISK, which 
promptly found two lost cluster sets. I saved them and looked at the contents; 
obviously these were clobbered versions of the report I had just abandoned. 
Multiplying the highest record number I could see therein times the record 
length yielded a file size well >300 MB, of which perhaps 40 MB had survived as 
file*.chk. Needless to say, this is quite prohibitive for Internet reporting 
purposes.

Results type B:

These were the attempts to rescue the installation by rerunning the installer. 
In both cases, the installer program simply ab-ended with a GPF before 
displaying any screens.

Variation #1: I looked at the GPF details screen. The fault was caused in my 
version of UNICOWS.DLL, which had in fact been installed by MS Windows Media 
Player. Due to system instability, my memory is again the sole resource here, 
meaning that I was absolutely unable to save any further particulars.

Variation #2: I didn't look at the GPF details screen this time.
Comment 1 Olaf Felka 2003-07-21 09:54:14 UTC
This is not a gebneral problem. I can't reproduce, OOo 1.1 is running
fine on my machine. Which version of unicows dll do you have?

*** This issue has been marked as a duplicate of 15288 ***
Comment 2 Olaf Felka 2003-07-21 09:54:29 UTC
Dupe.