On 2025/3/12 9:45, Gao Xiang wrote: > > > On 2025/3/12 06:55, NeilBrown wrote: >> On Mon, 10 Mar 2025, Gao Xiang wrote: >>> >>> - Your new api covers narrow cases compared to the existing >>> api, although all in-tree callers may be converted >>> properly, but it increases mental burden of all users. >>> And maybe complicate future potential users again which >>> really have to "check NULL elements in the middle of page >>> bulk allocating" again. >> >> I think that the current API adds a mental burden for most users. For >> most users, their code would be much cleaner if the interface accepted >> an uninitialised array with length, and were told how many pages had >> been stored in that array.> A (very) few users benefit from the complexity. >> So having two >> interfaces, one simple and one full-featured, makes sense.
Thanks for the above clear summarization. > > Ok, I think for this part, diferrent people has different > perference on API since there is no absolutely right and > wrong in the API design area. > > But I have no interest to follow this for now. Just to be clearer, as erofs seems to be able to be changed to use a simple interface, do you prefer to keep using the full-featured interface or change to use the simple interface? > > Thanks, > Gao Xiang
