Julian Foad <julian.f...@wandisco.com> writes: > However, this attempt at the patch is ugly. It stats the file to see if > it exists, and I just get the feeling it should be doing it some > "better" way. Something should be different about the structure of the > code - probably the larger tree conflict handling code and read-only bit > twiddling code outside the block that this patch touches - but I can't > see what exactly. > > How can it be better?
If it is valid for the file to not exist add a "may-not-exist" flag to the workqueue item, and when running the item set ignore_enoent=TRUE. -- Philip
