#134: Change MD5 hash construction to avoid false-positives #139
Conversation
Note that this will cause current transactions to become unique due to the change in MD5 hash construction.
Includes a new transaction in transfer.csv, and a small header for parsed_transfer.txt, which already wasn't the literal (ie. untouched) result of parsing transfer.csv.
It doesn't look like this is used for anything, and the tests don't fail without it, but correct me if I'm wrong ofc
A little trigger-happy with the force-pushes tonight.. EDIT: ..really trigger-happy, I just keeping seeing little things; but I figure no-one's pulled or merged or looked at it yet.
|
A little trigger-happy with the force pushes there; I think it's worth it for the clean commit history, and just keep thinking that's gonna be the last little tweak :c |
|
I would like to re-work this to be able to update hashes in-place, if only to learn more of python, as I don't know where to start with file editing and streams short of calling out to sed :p It would probably work best with another argument though, and I'm hesitant with that, least the handling of this depreciated footnote grow to take up more space than it ought to be welcome to. Detecting and updating old hashes could just as well be an additional helper script, which may be best if one would like to preserve the precedent of the ledger file going touched unless it's also used as the output file. Could even make the '--depreciated-md5' arg unnecessary,? Better argument names are always welcome. |
|
Re-submitted as #140 because I was naively working on master and, as it happens, Github doesn't let you change your default branch, only rename it. |
#134
Note that this will cause current transactions to become unique due to the change in MD5 hash construction,
which may be mitigated by augmenting
--skip-dupesor--confirm-dupeswith--depreciated-md5.