On Wed, Sep 12, 2018 at 01:08:35PM +0200, Alexander Bluhm wrote: > On Wed, Sep 12, 2018 at 09:01:41AM +0200, Claudio Jeker wrote: > > Why u_char instead of int? The errors in that list are used in errno which > > is an int. So my usual feeling is to use the type which is naturally used > > and don't try to optimize for space. > > Basically I want to have it the same for inet and inet6. The numbers > fit, FreeBSD uses u_char for both, so why not save some memory?
My main concern is that some day someone adds a errno that is > 255 and causes an overflow. Probability for that is close to 0 but I start to don't like these optimisations that are gone shoot you in the foot later on. > I have already commited it with mpi@'s ok. I don't insist on u_char, > you can change it to int if you want. Have seen it. Will think about it. -- :wq Claudio
