On Thu, Jul 30, 2020 at 17:27:35 +0300, Nikolay Shirokovskiy wrote: > Most of bitmap setBit/clearBit/getBit users know that the bitmap index is > not out of bound and thus don't check the return value. More precisely > the stats is next: > > Method all check > =================================== > virBitmapSetBit 85 14 > virBitmapClearBit 15 3 > virBitmapGetBit 15 6 > > where 'all' is the number of all occurences of the method and 'check' is the > number of occurences with 'if (method' pattern. > > Thus keeping the retvalue checking requirement produces more > noise then helps. I guess we even can make these function return > void as users can simply compare the index with the bitmap size. Well. An ignore_value is not really expensive and it makes the callers aware that something needs to be checked. I don't really see the point of this. Additionally, individual patches are missing justification in the commit message. Mentioning it in the cover letter is not enough as that doesn't get comitted.
