hashcash-0.32 - 09-Apr-2004 - Adam Back <adam@cypherspace.org>
* documentation fixes
* change multiple regexp behavior; previous algorithm only allowed
higher overrides; need to support both higher and lower
overrides. This also required introducing -o option to join
regexps which are set intersections where otherwise risk of
uninteded override occuring and mail being rejected as spent or
insufficent bits. Now revert to lexical order most specific
regexp first.
* wrote test script test.sh
* fix a few minor bugs uncovered by above test script
* -c now means check date
* allow -n etc with -X
* introduced -b relative to default way of specifying bits
* -b is no optional, if want token fully checked, but can give -b
default; or new relative to default -b +0.
hashcash-0.31 - 01-Apr-2004 - Adam Back <adam@cypherspace.org>
* final 0.x version (v0 format) release before 1.x version (v1
format) (bug fixes / maintenance only afterwards on 0.x version)
* remove -O3 from Makefile, use -O instead as fails on HPUX or
sun.
* fix some out of date usage stuff in hashcash man page.
* disable timing loop unless timing needed
* fix multiple reciept bug in -cX/-cx reported by Junior Ang
<junior@chrysant.com>. If you receive a mail multiple times
because you are on the receipt list multiple times, there will
be multiple hashcash headers for you. In this case it is
necessary to examine the first matching, non-spent stamp. The
bug was previous versions stopped on the first matching stamp
and then failed because it was spent. Need to keep going and
check later also matching stamps until find one which is not
spent.
* rationalize command line args further. No implied -m , more
things that are awkward to implement but not that useful are
disallowed.
* change purge operation to use read-write operations in the same
sdb file rather than creating a temporary file. This makes
locking easier and is also aesthetically nicer.
* add flock(2) database file write locking, and change creation
logic to use open(2) to avoid creation db race-condition also.
* make resource string case insensitive by default to match email
semantics; add -C option to force case sensitivity if desired
(email addresses are converted to and stored in lower case, so
you have to both mint and verify with case sensitivity turned on
to make use of case sensitivity)
* support minting multiple resources with multiple command line
args. Also if no resources given on command line, read
resources from stdin.
* support supplying multiple email addresses, for people who want
to accept as multiple addresses.
* support multiple resources on purging also.
* support multiple tokens with check mode as cmd line args, if
none given as args, read tokens from stdin; if -X/-x read from
cmd line args, then from stdin as email (matching stamp headers
skipping stamp headers)
* rename default simple database to hashcash.sdb (.sdb extension),
to distinguish from planned support for better database.
* fix bug in PPUTS didn't match PPRINTF
* fixup -l, -w, -n so they support multiple tokens also
* made use of -b optional (get the default on mint & check)
* added "-b default" to specify default number of bits with -s
(otherwise no way to measure the default speed without
specifying the number of bits -- and when this can change over
time it would be inconvenient for scripting to have to
separately obtain this)
* added support for wildcard email addresses with '*' wildcard
marker. '*' before '@' does not match '@', '*' after '@' does
not match '.'. And both email addresses must contain @ sign and
same number of '.' separated sub domains as wildcard address.
Wildcard matching is the new default. Use -S to get plain
string match. Can turn back on with -W.
* increased size of random string to reduce chance of collisions
between users. Now negligible chance of collision with typical
token sizes.
* added support for regexps. Can work from POSIX library or BSD
regexp library. Use -E to get regexps. Input is always in
POSIX syntax (specials are not quoted to have special action;
are quoted to have plain meaning). If using BSD library still
give input in POSIX syntax, it's converted to BSD internally.
* implement highest matching semantics. Ensures that eg -c -b10
*@bar.invalid -b15 adam@bar.invalid will not accept a 10 bit
token for adam@bar.invalid. (This is done by sorting resources
highest bits required first and accepting only the first highest
matching resource.)
* change arg parsing so -b, -e, -g, -z, -E, -W, -S, apply to the
following resources and tokens, and can be changed for later
resources/tokens with tokens and args interspersed. Means you
have to give these args before the resource/token or you will
get defaults.
1a48a554a7