Today, we bring you the first installment in a series of short interviews with lead/prominent developers of many "smaller" operating systems. In this new series, dubbed "Five Questions", every interviewee will answer the same five questions about the project they are part of. The series will be kicked off by Axel Dorfler, the prominent Haiku developer.
To read all comments associated with this story, please click here.
Member since:2006-12-01
I think they should test on as much hardware as possible.
BTW developing using a Virtual Machine is correct IMHO; The VM provides a sort of generic HW, and many developer can focus on developing new features and fixing bugs without having to deals with drivers or strange hardware.
I know hardware support is critical, extremely critical. But I think it would be really important to have an full working os and after dealing with different kind of hardware.
It may sound like chicken and egg problem because you cannot have a working os without hw support.. but really if the programming resources are not infinite you must choose what's more important.
my 2Cents.
Member since:2005-07-07
To answer the sixth question: that's the plan. And while using emulators is great for testing and developing, it's not a substitute for the real thing.
In fact, I have a Core 2 Duo, a Pentium 820D, a Pentium-M, a dual PIII, an Athlon XP, a PIII, and a standard P4 (with "hyper threading") to test Haiku with. Despite current AMD hardware, it's a pretty extensive test park for a single room.
Needless to say, Haiku runs on all of these machines more or less fine, also with SMP enabled. When getting closer to the first release, we'll try to make sure it will run on as many hardware as possible.
Member since:2007-03-18
Member since:
2006-11-30
I have a sixth question: Will Haiku even boot on modern machines anytime soon? I have three fast multicore laptops (amd and intel), and Haiku doesn't boot on any of them. According to the good folks in #haiku, there are either issues with smp, or something timing related on fast machines. Even when I disable smp, the darn thing won't run for long.
Now, that's fair... this is pre-alpha stuff. However, my concern is that most devs seem to do most of their development on emulators. As comfortable as that might be, it is a catastrophic practice. It means all devs work mainly on the exact same "hardware", and tons of chipset specific bugs, etc doesn't show up. And even the core devs, including axeld apparently, doesn't even have proper hardware to test things on (no 64-bit smp, etc)
Could someone please send these guys some test hardware? Not that I think they even ask (they should - other os projects have been successful in that)