On 10 April 2010 11:45, Daniel Trebbien <dtrebbien@gmail.com> wrote:
one, minor suggestion: why not name `end_message` simply `end`? I am of the opinion that naming should follow the language's standard library conventions closely. And, as the STL uses short, concise member function names, it is best to mimic this.
I picked that one thinking of the ASCII control codes END OF TEXT/TRANSMISSION/MEDIUM and the email convention of <eom>. And I realize it's not a member function, but STL isn't afraid of calling something lexicographical_compare if it needs to. As part of the http://en.wikipedia.org/wiki/Merkle%E2%80%93Damg%C3%A5rd_construction it's called "Finalization", and most precedents seem to use something like that. Crypto++ calls it "Final", the MD5 reference implementation uses "MD5Final", and the implementation provided for the "best understood" (according to NIST) SHA-3 candidate CubeHash (http://cubehash.cr.yp.to/simple.c) also calls it Final. To get an actual verb, I guess the most logical name would be "finalize()", but I suppose most libraries can't use that since their GC stole the name.