On Thu, 30 Jul 2009, Szak�ts Viktor wrote: >> The above behavior describes problem with nonblocking socket so it's >> expected for Harbour versions before: >> 2009-07-29 05:19 UTC+0200 Przemyslaw Czerpak (druzus/at/priv.onet.pl) >> With today CVS it should work without any problems. >> Are you sure that you made clean Harbour build before you make the tests? >> Few minutes ago I compiled httpsrv in my Linux box in four versions >> (native linux binaries with and without hb_inet* and windows MinGW >> binaries with and without hb_inet*) and tested them (Windows binaries >> executed using WINE). All of them work as expected and I cannot replicate >> your problem. Above suggests for me that for some reasons you may not use >> cleanly compiled Harbour code. >> Maybe it's a problem with hbmk2 paths so some old libraries are used or >> sth like that (i.e. old harbour.dll in shared build)? > I had similar problems on darwin until I did a full rebuild.
Exactly. AFAIR Darwin inherited kernel from one of BSD systems so unlike Linux it should exploit the problem caused by inheriting O_NONBLOCK flag in handles returned by accept(). WINE emulates such behavior in Linux so I guess it's also in Windows and it's the reason of Francesco's problems. I really think that using clean Harbour build to compile and link httpsrv should resolve them. best regards, Przemek _______________________________________________ Harbour mailing list Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour