On 4 September 2014 20:29, Jan Vesely <[email protected]> wrote: > On Thu, 2014-09-04 at 16:05 +0200, Martin Decky wrote: > > > I think checking the queue pointer for null and refusing transfer would > > > be the right solution here. > > > > Why not just keep the workaround there, even if it goes beyond the > > specification? I mean, it is the classical robustness principle: Be > > conservative in what you send, be liberal in what you accept. > > I don't see how it would improve robustness more than checking for NULL > queue. I disagree with the application of that principle in this case. > Being liberal hides bugs, that's ok in case of hw or third party broken > sw (it can't be changed and such workarounds should be clearly marked), > but sw that is under our control should be fixed. > > I think that makes sense as long as the actual problem is fixed. I'd much rather have a workaround in the code than a nonfunctional thumb drive. Do we even know whether it's a QEMU bug or just funky hardware?
-- Jirka Z.
_______________________________________________ HelenOS-devel mailing list [email protected] http://lists.modry.cz/listinfo/helenos-devel
