Your probably right about sorting filespecs/patterns but that would entail a fairly large change in the logic of the dll (and a lot of rewriting/debugging) . The actual time taken could be at least halved if the matching was only done once (it's done at least twice that I know of - once to add up sizes and once to actually extract the file but it might be done again to see if the pattern matches anything [I have not found this bit yet]), but again this would entail many changes. As I said I am working on a windows-like change in pattern matching, so I am open for any idea's as to how it should work. - Russell ----- Original Message ----- From: "Angus Johnson" <ajohnson@xxxxxxxxxx> To: <delphizip@xxxxxxxxxxxxx> Sent: Thursday, March 28, 2002 8:49 AM Subject: [delphizip] Re: At last ... (hopefully with the file) > Russell, > > > That's right and correct (as per the original) > > OK, I haven't looked closely at the original code, but if this is the case > then I believe that this behaviour should be corrected. > > Well done on what you've done so far though. > > By the way, I'm still confident that separating out non-wildcard patterns, > and handling them differently (by sorting both them and the comparing > strings > prior to matching) would also make a big difference in speed. > > > > using - apath\*books.doc - as the pattern > > > would your code not also find - apath\anyname.doc - as a match > > Angus > > >