On Sun, Mar 4, 2018 at 3:14 PM, <lars.schneider@xxxxxxxxxxxx> wrote: > Changes since v8: [...] > > Thanks a lot Eric for your great review! I think I fixed everything you > objected with one exception. You noticed that the current code only > checks for BOMs corresponding to the declared size (16 or 32 bits) [1]. > I understand your point of view and I agree that any BOM in these cases > is *most likely* an error. However, according to the spec it might > still be valid. The comments on my related question on StackOverflow > seem to support that view [2]. Therefore, I would like to leave it as > it is in this series. If it turns out to be a problem in practice, then > I am happy to change it later. OK for you? Fine. As this is not a correctness issue with the conversion itself, but rather an attempt to detect misconfiguration by the user, it can always be revisited later if someone comes up with a more solid argument that one interpretation is more correct than the other. I did leave a few review comments on v9, but I don't think any are necessarily worth a re-roll. If anything, and if they seem actionable, then perhaps a patch or two atop the series could address them (at some point).
