Hi Avraham,

Avraham Hanadari wrote:
I know how to remove the user data. I have done it several times, and 
afterwards the PDF function works. I still want to know where in those user 
files the setting is that enables or dis-enables the PDF function. None of the 
OOo experts who have kindly suggested solutions over the past couple of years 
seems able to tell me. OOo grew into this snaggle. Perhaps it will grow out of 
it as well.

Could you let me see the whole folder of the broken user settings if you still 
have it?

I am suspecting a racing condition. OpenOffice.org is running enforced with the 
multi-thread technology. User can start saving a 100s slide Impress document to 
a file and it takes several seconds. Without waiting for OpenOffice.org to 
finish the task, user can switch to another application like Writer or another 
Impress document and work on it.

Files in the folder for the user settings data are shared with all instances of 
applications and would be repeatedly updated at the several events: opening a 
document file, saving a document, moving a location of window, pressing the OK 
button in some dialog windows, ...

Several threads can access to the user setting data through the single module 
called configuration manager. If two threads simultaneously make their own 
request to the configuration manager to update a single file - one file can 
holds several different items -, those requests should be processed properly. 
The configuration manager must be designed to work under such a racing 
condition.

We know there is nothing perfect in the world.

The configuration manger is designed platform-independently. It runs above the 
System Abstraction Layer (SAL) of OpenOffice.org. Hiding differences among 
individual OSs, SAL helps module programmers write their own source codes 
platform-independently. That brings us lots of benefits and at the same time it 
might lead a few problems such as a platform-dependent issue.

Problems about broken user settings data have been reported for years, as you 
have been pointing. I am not sure about the reasons why Windows versions of 
OpenOffice.org seem to likely experience this type of problems than other 
platforms. Is it because majority of users are using Windows? Is it because 
writing source codes with multi-thread for Windows requires something special?

Although we might succeed identifying the exact location of broken data in the 
exact file out of the user settings data files, which would not directly help 
to identify the real cause. If it was easy, the problem had been already solved.

The problem unlikely occurs or can not be reproduced in a clean laboratory 
where programmers are working while it seems that the problem likely occurs in 
your place. That would help software engineers to investigate the problem and 
eventually solve it.

Tora

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to