> Then I restart the program with the option 'restart an 
> interrupted mirror'. And after some time I interrupt the 
> program again, in the same way as previously (i.e. 
waiting 
> a huge amount of time until the program ends 
> finishing/closing/interrupting the project).

Well, in this case, this is normal: the previous cache is 
used only when files are needed, and the new cache is 
feeded with both network and cache data progressively. 
There is no "brutal" merge between the two caches, even 
when interrupting the engine.

It means that if you interrupt the mirror in the begining, 
the new cache will only contains what have been "scanned" 
during this session. It may look weird, but remember that 
you can change any option in httrack between two sessions, 
including URLs: what if you change the previous URL to 
another one? The previous cache data will be irrelevant, 
and keeping the data would be overkill.

The only workaround could be to do a merge of "data not yet 
seen in cache 1 but seen in cache 2" when interrupting a 
mirror - but the best way is maybe to delete the new cache 
(new.* files) and let httrack use the previous one.