Re: [Babeldoc-devel] Scanner changes...

[Bruce M. <br...@mc...>]

On Friday 11 April 2003 06:34 am, Dejan Krsmanovic wrote:
> Hi all,
> 1. I am planning to do few refactorings in the scanner module. I am not
> sure if some of these should be left after 1.0 release but this is what I
> plan to do:
> - Renaming few classes. Personaly I think that using term "message" for
> documents that are found using scanner leads to confusion. So I suggest
> replacing all occurancies of 'message' to 'document'.  For example:
> ScanMessage -> ScanDocument
> ScanMessages -> ScanDocuments
> IScanMessages -> IScanDocuments

No, make these changes now - before 1.0.  There are pretty simple and will be 
no problem to sort out before 1.0.

> Would this change cause problems? Should I leave original classes but
> depracate them (Altough I don't see a benefit since users will probably use
> only ScannerWorker class).

No - replace them - they are contained within the scanner module.

> 2. I am also moving most of functionality from ScannerThread to
> ScannerWorker class. ScannerWorker will perform scanning and feeding the
> pipeline. It will be abstract class (as it is now) and subclasses will
> implement few methods for scanning specific resources. Subclassing
> ScannerWorker should remain the same.
> Moving logic to ScannerWorker will enable us to use different sheduling
> methods for the worker. I plan to use Timer and TimerTask for scheduling
> instead of thread. This will disable paralel working of two scanners since
> TimerTasks are executed one by one in the same thread. I also plan to add
> some methods for management purpose to prevent deadlocks. I am not yet sure
> how this will be implemented but some kind of timeout period should be
> defined and another thread shoud check each worker if timeout period has
> elapsed and kill it if it has...

Bigger change - lots of testing here...

> 3. I was also thinking of using seconds instead of milliseconds for period
> option in scanner configuration. Does anybody use scaning period less then
> 1 second? I guess no. This is also more user - friendly, we should not
> force users to think as us - developers. I guess milisecods was used in the
> first place because Thread use miliseconds in its methods (Am I right,
> Bruce?)

I dont know about this one.  I think that at this point only highly technical 
users are using babeldoc and therefore should be able to grasp the concept of 
milliseconds.  In post 1.0 we will have nice GUIs so that we can simplify 
things for managers and secretaries...  :)

> Since Babeldoc 1.0 should be released soon, I think 2. should be left for
> versions after 1.0.
> 1. and 3. could be changed now.

1 definitely - 3 I dont know about.

> Any opinions?
>
> Dejan