Original post: hachyderm.io (Mastodon)
So we need to be careful with upper- and lowercase. Meanwhile the docs: > settiings
Yes, the settiings are different than the settings. You also need to be careful with those.
had to use a different spelliings at backend and frontend, otherwise it wouldn’t work.
no, settings = settings but settings != Settings, as we all know.
There’s a double-i in “settings” in the documentation screenshot
They specifically said “this is not a typo”!!!
Is the backend Python and the frontend JavaScript? Because then that would happen and just be normal, because Boolean true is
Truein python.Probably, but if you’re interpreting user inputs as raw code, you’ve got much much worse problems going on, lol.
[...]®ister=import os; os.system("sudo rm -rf /"); return TrueHey, that’s my username too. Or it was going to be, while the site was still up.
What a coincidence!
I guess I’ll wait for the site to come back, and see if it’s still available…
It’s the settiings file… It’s probably supposed to only be written by the system admin.
A good place to put persistent malware. That’s why when using docker images always mount as ro if at all possible.
It’s you can modify the settings file you sure as hell can put the malware anywhere you want
It’s you can modify the settings file you sure as hell can put the malware anywhere you want
True. (But in case it amuses you or others reading along:) But a code settings file still carries it’s own special risk, as an executable file, in a predictable place, that gets run regularly.
An executable settings file is particularly nice for the attacker, as it’s a great place to ensure that any injected code gets executed without much effort.
In particular, if an attacker can force a reboot, they know the settings file will get read reasonably early during the start-up process.
So a settings file that’s written in code can be useful for an attacker who can write to the disk (like through a poorly secured upload prompt), but doesn’t have full shell access yet.
They will typically upload a reverse shell, and use a line added to settings to ensure the reverse shell gets executed and starts listening for connections.
Edit (because it may also amuse anyone reading along): The same attack can be accomplished with a JSON or YAML settings file, but it relies on the JSON or YAML interpreter having a known critical security flaw. Thankfully most of them don’t usually have one, most of the time, if they’re kept up to date.
I curse the sadist who decided True should be uppercase in Python
guido, why did you make python so weird?
In this instance, I think there was some suggestion to write code in mostly lower case, including all user variables, or at least inCamelCaseLikeThis with a leading lower case letter, and so to make True and False stand out, they’ve got to be capitalised.
I mean. They could have been TRUE and FALSE. Would that have been preferable? Or how about a slightly more Pythonic style: __true__ and __false__
Can’t they just convert a “true” input to backend to uppercase
Depends on how it’s set up. If the setting is going into the env it’s a string, so I’d expect some sort of
if os.getenv("this_variable", "false").lower() == "true": # or maybe "in true, yes, on, 1" if you want to be weird like yaml this_variable = True else: this_variable = FalseExcept maybe a little more elegant and not typed on my phone.
But if the instructions are telling the user to edit the settings directly, like where I wrote this_variable=True, they’d need to case it correctly there.
Fyi, using a condition to assign a boolean is equivalent to assigning the condition itself. No need for the IF.
true, though sometimes i find the more verbose style easier to read, and more maintainable (eg: you want to do something else in the block, you can just add a line instead of changing your ternary / etc). Small things
Could be worse. At least it’s documented
To me the biggest problem, that AI might solve is documentation.
Have you tried to use AI for documentation? It’s pretty shit.
Implying Hell is frontend… yeah, actually, that tracks.
Glorious. I remember some hilarious nonsense in an API where the devs I worked with hadn’t known they could just use boolean in JSON and had badly implemented it through strings, but this… This is amazing!
At my last job we had a lot of old code, and our supposedly smartest framework people couldn’t be bothered learning front end properly. So there was a mix of methods for passing values to the front end, but nobody seemed to think of just passing JSON and parsing it into a single source of truth. There was so much digging for data in hidden columns of nested HTML tables, and you never knew if booleans would be “true”, “TRUE”, “1”, or “Y” strings.
Never mind having to unformat currency strings to check the value then format them back to strings after updating values.
I fixed this stuff when I could, but it was half baked into the custom framework.
A system I work with gives all keys a string value of “Not_set” when the key is intended to be unset. The team decided to put this in because of a connection with a different, legacy system, whose developers (somehow) could not distinguish between a key being missing or being present but with a null value. So now every team that integrates with this system has to deal with these unset values.
Of course, it’s up to individual developers to never forget to set a key to “Not_Set”. Also, they forgot to standardise capitalisation and such so there are all sorts of variations “NOT_SET”, “Not_set”, “NotSet”, etc. floating around the API responses. Also null is still a possible value you need to handle as well, though what it means is context dependent (usually it means someone fucked up).
