On samedi, f=E9v 8, 2003, at 19:41 Europe/Paris, abel deuring wrote: [ snipped useless-and-out-of-topic rant ]
> I also had a short look into Apple's mailing list archives. The=20 > answers to a question by Beat regarding device access from forked=20 > processes seem to indicate there is really not hope. The main reason=20= > seems to be "too complicated to implement for Mach". (BTW, the=20 > archives of older mailings are the worst organized archives I have=20 > seen since a long time. So much for the shiny surface of Apple=20 > stuff...) [...] > Somebody proposed to the apple mailing list to start a sort of a=20 > daemon which accesses the device. Since sockets (pipes too?) seem to=20= > survive a fork on macos, this could indeed be a way to avoid the=20 > current fork hassles. Other ways would be to provide a non-forking=20 > version of the backend (like the Sharp and NEC backends do), or to=20 > make the backends thread-safe (should not be that much work, since the=20= > tasks of the two processes/threads are different enough), and to=20 > expand the sanei_thread library. I'm not really an IOKit expert, but perhaps should one write a generic=20= device driver for USB that just create proper device entries. I suggest emulating *BSD USB behaviour. That way, we then would need to=20= change libusb to use *BSD code and we would have a better=20 compatibility. IOKit use as its advantage, like allowing USB access in=20= userland, but the mach_port problem is not a simple issue. Just a thought, I'll dig further to see what can be done (not sure I'll=20= have the time to actually code anything) Hub --=20 Hubert Figui=E8re - http://www.figuiere.net/hub/ Cell-phone: +33 6 18 01 42 11 - iChat/AIM: hfiguiere Any HTML e-mail sent to me will be discarded.
