On 2014-01-11, Gary Gregory wrote: > The reports look good BUT I think the NEW code should have better code > coverage. There seems to be some non-trivial code that is not tested > at all.
AFAICS this is mostly handling cases that are difficult to create test data for. * Snappy with literals longer than 64kB or offsets larger than 64kB. Either will never happen using the framed format as blocks are limited to 32k there. * .Z without block mode - no idea how to create such a stream, compress doesn't have a command line option that would control it. I'd assume such streams don't really exist. * .Z and clearing Unix compress will clear the LZ78 table when it detects compression ratio going down. It would take quite a bit of experimenting to create such a test file. * Unshrinking ZIP and (partial) clear This is even worse as you need PKZIP < 1.1 to even create said archives - and would require the same amount of experimenting to get there. Stefan --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
