Andreas Färber <andreas.faerber@xxxxxx> wrote: > Of course there can be exceptions to this, where a project doesn't > stick to such best practice rules, but I'm pretty sure most "high > quality" projects don't want to have new #defines introduced for > Haiku > with no evident benefit. Sure, but for example networking code heavily needed customizations for BeOS - this will no longer be the case for Haiku; in that case it's better to specifically check for BeOS, or against Haiku. > The only tricky thing I've encountered was in apr(-util?) where BEOS > was used to conditionally define a prototype for strerror_r because > it > was implemented in libroot but not present in their headers. There > I'll have to look into a new define, but I don't see the point in > replacing BEOS with BEOS || HAIKU, it's much easier to define both > and > only use #if defined(BEOS) && !defined(HAIKU) or similar only where > needed. Yes, and Haiku's GCC still defines both __BEOS__, and __HAIKU__ for that reason. > We can still do a full code review once we have the self-hosting set > of tools working on Haiku. If we don't actually port the tools, we could also just use BeOS binaries, though :-) Bye, Axel.