[Box Backup-dev] Re: [Box Backup] Win32 port

Gary boxbackup-dev at fluffy.co.uk
Tue Dec 6 18:02:08 GMT 2005


Chris, Ben, Nick,

> Nick made some win32 changes to your code, which I've 
> collected as a small patch, attached.

Oh, I see now, so the change pathway would be as follows:

0.9 >> [Nick]: Win32 patches >> 0.9-win32
0.9 >> [Gary]: KeepAlive patches >> 0.9-mod
0.9-win32 >> [Nick]: KeepAlive patches >> 0.9-win32-mod
0.9-win32-mod >> [Nick]: KeepAlive port >> 0.9-win32-mod-2

Considering the nature of 0.9-win32-mod-2 and the overall code, I think
it would be best to unmerge the 0.9-win32-mod2 changes completely. The
question now is what to do with them.

I have to admit here that the 0.9-mod was more of a QFE than a
non-codebase intrusive change, but, on the other hand, if you look at
the 0.9 MaximumDiffingTime implementation (based on a static flag and
signal), it is not too pretty either, since it was not designed with
multiple interruption intervals (in other words, multiple diffing
process "observers") in mind.

I would be sad if the future version of BoxBackup did not include the
KeepAlive time/command - it would make the product unusable for me
(even with the excellent 0.9-chromi, KeepAlive is still needed for
diffing 4GB files, otherwise you diff for only 300 seconds and upload
1GB+ each time over possibly a slow upload link).

I think the way to go would be to:

1.) Modify vanilla 0.9 to better structure the observer code.
2.) Re-apply the KeepAlive >>command<< 0.9-mod parts.
3.) Manually add the KeepAlive >>observer<< to the (improved) 0.9 code.

This way we would be getting both the KeepAlive command and a clean
solution to the observer problem, which I am relatively sure will be
needed in the future anyway (think, say, diffing progress notifications
every once in a while).

Gary



		
__________________________________________ 
Yahoo! DSL – Something to write home about. 
Just $16.99/mo. or less. 
dsl.yahoo.com 




More information about the Boxbackup-dev mailing list