René Scharfe <l.s.r@xxxxxx> writes: > ... But > a replacement for git_mkstemp_mode() with two umask(2) calls looks less > attractive to me than fortifying git_mkstemps_mode() with a good source > of randomness. True. Also, it is not like we are supplying our own implementation of random source, but are just pluggig various system-supplied random source into our code, so I do not see the "auditatiblity" problem we heard earlier too much of an issue.
