Re: diff compression

Hi,

> Just a status to let you know.
> 
> I have started implementing an adhoc interface to librsync into libdar
2.5.x however I face what seems to be a bug in librsync which I have no time
to troubleshoot nor fix.

Whoa...thanks for the update :)

> more details at
> 	https://sourceforge.net/p/librsync/mailman/message/34205320/
>

I have checked your program and the message on SF. I have done a quick
review of the code (I apologize if I'm stating something obvious...let me
know :-))
The test program crashes because tag_table is NULL:

(gdb) print sig->tag_table
$3 = (int *) 0x0

Tag_table is initialized in rs_build_hash_table()...
Now, rdiff program calls rs_build_hash_table(sumset) explicitly and then
uses rs_delta_file() which in turn calls rs_delta_begin(). Your 'my_rdiff'
only calls rs_delta_begin()...

Adding rs_build_hash_table(sumset); before rs_delta_begin() no longer
crashes the program and I don't see rs_build_hash_table() being called
anywhere from librsync, so I guess the table has to be built explicitly...

> except that point, the base relation between librsync and libdar is done.
Remains a lot of work to do for within libdar.
> 
> Due to the lack of time and already 6 month since the expected release of
2.5.0 I must postpone this feature to 2.6.x this feature.
> However I will start by its implementation, so maybe a new branch could be
done sooner than the average 3 years between major releases of libdar.
> 
> 
> Thanks for your comprehension,

Thank you for the update and keep us informed :-)

Regards,
gulikoza



------------------------------------------------------------------------------