Created by [CecilWesterhof]. **Getting an UUID on a *NIX System** There are some problems with getting the [UUID] (type 4) on *NIX systems. For one they are not really random. Because of this I created the following function: ====== proc getUUIDNix {{secure False}} { set fortyeightBits [expr {2 ** 48 - 1}] set sixteenBits [expr {2 ** 16 - 1}] set thirtytwoBits [expr {2 ** 32 - 1}] set twelfBits [expr {2 ** 12 - 1}] format %.8x-%.4x-4%.3x-%x%.3x-%.12x \ [getRandomIntInRangeNix 0 ${thirtytwoBits}] \ [getRandomIntInRangeNix 0 ${sixteenBits}] \ [getRandomIntInRangeNix 0 ${twelfBits}] \ [getRandomIntInRangeNix 8 11] \ [getRandomIntInRangeNix 0 ${twelfBits}] \ [getRandomIntInRangeNix 0 ${fortyeightBits} False True] } ====== It uses getRandomIntInRangeNix from [Random Integers]. '''A Better way''' The above version is a straight implementation. But a more efficient implementation is: ====== # An UUID is built from 5 hex strings connected by a '-'. # Their lengths are 8, 4, 4, 4 and 12. # With version 4 the folowing is necessary: # - The first digit from the third string is a 4 # - The first digit from the fourth string is 8, 9, A or B. # The five random hex strings are generated. # The first digits of the third and fourth strings are changed. # The UUID is build from the five strings and returned. # Because binary scan returns lowercase letters toupper is used. proc getUUIDNix {} { binary scan [getRandomBytesNix 16] H8H4H4H4H12 hex1 hex2 hex3 hex4 hex5 set hex3 [string replace ${hex3} 0 0 4] set oldVal [scan [string index ${hex4} 0] %x] set newVal [format %X [expr {(${oldVal} & 3) | 8}]] set hex4 [string replace ${hex4} 0 0 ${newVal}] string toupper ${hex1}-${hex2}-${hex3}-${hex4}-${hex5} } ====== This uses getRandomBytesNix from [Get Random Bytes on *NIX]. ---- As always: comments, tips and questions are appreciated. ** See also ** * [UUID] for other implementations ** Discussion ** <>Linux | Utilities