BTW: ( @PulkoMandy ) Yesterday I created new patches for Termios in FPC and uploaded them to the bugreport. I am assuming it’s OK for now, though (as stated as comment in one of the files) I see a discrepancy between two implementations of a routine (leaving old one there for now).
I tested it, compiling our company’s app on Windows 7 with FPC 3.2.2 and LazarusIDE 2.0.12, for both 32 and 64bit. The linker these days works a big lot better than on the older FPC. Also I can compile 32bit, which (I think I remember) did not work before.
I installed the apps on both a 32bit and 64bit Haiku (around beta3). On 64bit, serial ports work OK, still apart from the flush command: would be nice if we can figure that out someday; I have it disabled still in Synaser therefore).
On 32bit, the serial ports just don’t work. With Haiku’s serialconnect app it’s OK, but not with lazarus. Though I -did- get correct reception into Haiku on one (random) test occasion (tx remained a problem).
On both systems usb_serial is complaining about not being able to set some interrupt or so. Don’t know how that was before, I did not check the syslog back then.
About the patches, since noone responded to my request for help, I decided to just go forward as good as I can.
Oh EDIT: the flush: at least I am wondering if it’s good practice to abort the flush if there’s no ‘other side’ available. One could have also decided to just abort trying to flush the ‘other side;’ but -still- go ahead and clear ‘this’ side, when IOFLUSH is requested that is. On OFLUSH only, an error is OK of course…
EDIT2: Oh yes, I saw that message indeed. Unfortunately getting the recipe to compile is already not something I can pull off apparantly, so I am not spending time on this atm. (Still I would love it if gdb would work… )