| Age | Commit message (Collapse) | Author |
|---|
|
a preprocessor macro
Fixed issue with tcoin code1 and tcoin code2. Now they're sent via stdin instead of as arguments to tcoin.
Refactored file permissions into a preprocessor macro.
|
|
TCOIN_PASS_RESET_CONTACT_EMAIL to parameterise contact details
Parameterised contact details via TCOIN_ERR_CONTACT_EMAIL and TCOIN_PASS_RESET_CONTACT_EMAIL will allow them to be customised for tilde.green's tcoin
|
|
Negative krowbar balances for pcoin too
|
|
As seems login forgot to close a parenthesis in the manual recovery
messages, and also appears like I am hunting for errors in tcoin :)
I have just changed the instances of the manual recovery messages from
"login@tilde.town (town-only) or login@tilde.team (internet wide [...]"
to:
"login@tilde.town (town-only) or login@tilde.team (internet wide) [...]"
|
|
I've tried to add try-catch blocks in those cases where fork()s fail.
It doesn't quite work well though.
|
|
pcoin now checks if an empty key was entered, and sends a different error message and error code for empty key as compared to incorrect key.
|
|
I can't believe I made this mistake, but I doublecounted minercoins, so now that doesn't happen.
|
|
Adding jmjl's 'ju coin' (which is a copy of tildebot) to the tcoin and pcoin balance
|
|
balance
pcoin -aib now doesn't allow you to add to your own internal balance. in the future, pcoin -ib <your own program name> will allow you to see what your 'net balance' is, i.e., pcoin -b minus pcoin -to
|
|
Thanks to ~jmjl, found a bug where pcoin's silentsend was sending the wrong message even though the send itself was correct.
|
|
Due to krowbar not being here and krowbar's tildebot not rejoining, shifting tildescores file location to jmjl's tildescores location. jmjl is running krobwar's tildebot with the name 'ju' in tilde.town's internal irc's #bots channel
|
|
This is expected to help a pcoin escrow program confirm that a key supplied to it is that of the right pcoin account by checking what the output of 'pcoin name' is. In the future, it may be expected that 'view key', 'spend key' and 'transfer key' would be three different keys that would split power over the pcoin account. And each of the previously mentioned 'keys' could be m-of-n keys. Maybe some kind of social recovery feature could also be added (where other users confirm who you are so you can get your account back, but you nominate them as your 'social recovery circle' beforehand.
|
|
old "~username" format
Made this change so that new "!minercoin ask" awarded minercoins will still show up in tcoin without losing old !minercoins awarded.
In other words, the inflation will continue until morale improves.
|
|
This will allow pcoin accounts to be transferred from one person to another
|
|
Thanks to ~jmjl, I was able to find that pcoin_keys didn't work. To fix it, tcoin uses sneaky magic but with stdin isntead of an argument, and a new library imported called popen2, that lets me send both stdin and receive stdout, to successfully get "ls <path to the pcoin_keys directory>" to work (it execl's another instance of the tcoin comand so that it runs under setuid instead of as the original user who ran it.
|
|
single file
This is preparing tcoin/pcoin for the automatic registration of program accounts.
Eventually, I want program accounts to be tradeable by there being a way to regenerate a new key.
So, send the current key to whomever you want the program account to be used by, then the new owner regenerates the key
to complete the transfer. The old owner will no longer be able to use the old key to operate the new account. The key is
long and the key generation random enough for key collisions to be virtually impossible.
|
|
Improved the way 5 arguments were handled, where the intention was to
send a message along with the amount and receiver, with a new error code
and a slightly modified error message.
|
|
Fixed a case of too many arguments in 'tcoin send' that wasn't handled.
Thank you for finding this ~julian.
|
|
There were many more places where I had hardcoded values, but
should have used strlen() or sizeof(). I have gone through
tcoin.cpp and pcoin.cpp, found those, and fixed them.
Hopefully, this is all of them.
|
|
In user_has_initialised(..), the path lengths were hardcoded.
This has now been changed to being calculated via sizeof().
In get_file_value and add_file_value, strlen() was used. This has
also been replaced with sizeof().
|
|
Refactored the send function in tcoin and pcoin, and in the process,
fixed a bug in pcoin's silentsend function.
|
|
On OpenBSD, the equivalent of %_d is %e. This is fixed for OpenBSD/tilde.institute type systems now.
|
|
1. Removed unneeded WHOAMI_PATH preprocessor constant from pcoin.cpp.
2. Updated WHOAMI_PATH for tilde.institute to use "id -ru" to get the real uid on OpenBSD (/usr/bin/whoami doesn't get the real username on OpenBSD).
|
|
And I don't know how to clean my commits after having pushed
them to remote origin/master. Maybe a Git stash and hard reset?
|
|
TCOIN_MSG_LENGTH_LIMIT
I made a switch from TCOIN_MSG_LENGTH to TCOIN_MSG_LENGTH_LIMIT for the symbolic constant
representing the maximum number of characters of a transaction message. I changed all instances
of the symbolic constant to this new name in tcoin.cpp, but forgot to do it in one instance in
pcoin.cpp. This goes on to show that I need to develop tests for tcoin and pcoin, as well as
start using getopt from util-linux for proper options handling.
|
|
After having received a 10000-character transaction message from ~jan6@envs.net,
I finally came around to implementing character limits on transaction messages.
Previously, I was of the opinion that one should be able to send a movie as a
transaction message if one so wished.
I took inspiration for an appropriate character limit from Australia's NPP (New
Payments Platform), which has a 280-character limit on transaction messages. I
strongly suspect this limit came from Twitter's analysis of tweets in different
languages and Twitter research that concluded that a 280-character tweet was the
appropriate length of tweet. A transaction message is supposedly like a tweet with
some money sent along with it.
|
|
I don't know how I missed this, because I remember coming across this deficiency before.
I had missed a case, where silentsend is requested but with a message. Previously, it would
fail without any message to stdout or stderr (only a return code).
Now, silentsend with a message works even on pcoin.cpp (tcoin.cpp already had this change).
|
|
I made a mistake. I should have done this before making my commits.
That's the only reason why this merge is necessary.
|
|
constants
I had forgotten to handle the fatal errors and take the error numbers out into symbolic constants. I have now done that in tcoin.cpp and pcoin.cpp
|
|
I had forgotten to change \u23CE to <new-line> and introduce a check for <vertical-tab> in many many places
in tcoin.cpp and pcoin.cpp. I should probably put these things in some kind of function so I don't repeat myself.
I'll probably do that in a future commit.
|
|
This doesn't change functionality. It is simply a refactor that separates error codes out into symbolic constants in tcoin.cpp and pcoin.cpp.
|
|
<vertical-tab>
Newlines and vertical tabs in messages are replaced with <new-line> and <vertical-tab> to prevent transaction spoofing inside a message.
|
|
The title is confusing, sorry. "No messages found" is shown in tcoin and pcoin when no messages are found (and this no-message message has one newline above and one newline below it instead of two newlines above it as before).
|
|
Several changes were made that were difficult to separate into separate commits. Since features are directly committed to master, there are no feature branches.
The high-level list of changes is as follows:
1. Fixed a bug in pcoin where using "pcoin send -s" for silentsend would forget to add the transaction message that provides a record of who transferred how much and when, even if there is no custom message appended (send_message(.) was not called).
2. Improved the help text to make it clear that sending messages could be done using silentsend too.
3. In several cases in tcoin and pcoin, certain ways of doing a silentsend would still send with verbose output because, by mistake, the option passed to send_message(.) was not changed from "verbose" to "silent".
4. In pcoin, there was the case of 6 arguments missing, where one does pcoin send -s <user> <amount> <message>. This case was handled in tcoin but not pcoin. Now, it is handled in pcoin too.
5. Now, when a custom message is sent in tcoin or pcoin, an additional text confirming the successful sending of the custom message as distinct from the one confirming the sending of the coins is displayed to the user.
|
|
The KROWBAR_OFF, DA_OFF and MINERCOIN_OFF symbolic constants are now included in tcoin_defs.cpp and pcoin_defs.cpp and have been removed from the main source files, tcoin.cpp and pcoin.cpp.
ntcoin has been modified to include #define KROWBAR_OFF, #define DA_OFF and #define MINERCOIN_OFF in pcoin_defs.cpp and tcoin_defs.cpp by default, since those three features are only (as of now) available on the tilde.town server and on no other server.
|
|
In pcoin.cpp, minercoin-score addition can be disabled by writing #define MINERCOIN_OFF near the symbolic-constant definitions.
However, in one place, a check for the absence of the DA_OFF symbolic constant was being made instead of MINERCOIN_OFF. This is now fixed.
|
|
I realised that different tilde servers will have different definitions but the same code otherwise, and 'git push' and 'git pull' to update the code on other servers would override those definitions.
Thus, the tcoin.cpp and pcoin.cpp files now look the same for all tildeservers and have a #include "tcoin_defs.cpp" and #include "pcoin_defs.cpp" respectively. The configuration, which is in the form of #define symbolic constants, is stored in tcoin/bin/tcoin_defs.cpp and tcoin/bin/pcoin_defs.cpp. ntcoin is modified to create these two files with the right definitions. ntcoin is also modified so that compile_tcoin and compile_pcoin now include the -I option, which instructs the compiler to look for includes in additional directories, in this case, tcoin/bin .
|
|
Benjamin Harris (https://ben.tild3.org, https://github.com/benharri, and https://benharr.is) found a bug where the time recorded for transaction logs/messages was local server time instead of UTC. It should have been UTC no matter what the local time was.
|
|
To fix the preprocessor constants so that they point to the right directory for the tcoin and pcoin executables. This will be different depending on where the tcoin and pcoin executables are stored.
|
|
Since not all places where tcoin runs have tildebot, minerbot and daily-adventure, these integrations are now disabled in the code through preprocessor directives (symbolic constants that act as flags)
|
|
pcoin.cpp and tcoin.cpp were missing the preprocessor directives generated by ntcoin. The right preprocessor directives are now included in both files.
|
|
pcoin.cpp had redundant code in 'int main()', where user_amount was redeclared (and thus became a local variable that shadowed the global version inside 'int main()' (!). Fixed by removing the redeclared, redundant line.
|
|
Added #define DEBUG and put a few check statements under it.
Fixed error message in pcoin that suggested the incorrect "add_internal_balance" amount when "internal_balance" owed to a particular user was not enough (and was lesser than the number of tildecoins being sent using the "send" command to said user). Also, fixed an error in retrieving the amount owed to a particular user when doing checks during the "send" command ("temp_program_sender_username" instead of "program_sender_username").
|
|
daily-adventure changed how it stores state, so this new code fixes daily-adventure integration to the new format that daily-adventure stores state in.
|
|
|
