hi,
i posted this already to de_bugs, assuming German is one of the developers native languages...
after aMule crashes, and, it crashes every 1 or 2 days and most times i let it run and run until it crashes... it leaves retrieved data and meta info about retrieved data in a mismatching state on disk:
-1- after a crash, each chunk which was busy downloading when aMule crashed, finishes as 'corrupted' when completed.
-2- i had multiple instances where files, when completed, were restored to download mode because the final checking found one or even two corrupted chunks. thats not only bad, its probably even worse, because i probably also served those broken chunks several times! i don't know whether the peers could detect the corruption, but at least i hope so.
so what to do?
please always write the chunk data to disk prior to writing/updating associated meta data to disk.
please checksum all part file chunks after (re-)starting aMule. Its really no good serving broken data.
besides that: good work!
... m2kio !