Re: [Feature Request] multiple files sélection in addition to password and "no file" agent

[devm23k73ju29h3r at dolce-energy.com]

Le 10/06/2026 à 23:57, Robert J. Hansen via Gnupg-devel a écrit :
>> the key here is the limit of human brain... remembering a pure random 
>> 12 char is too much for most people, me including
>
> Yes. So use one you can remember, along with an appropriate key 
> derivation function like Argon2, PBKDF2, or the one OpenPGP used prior 
> to the advent of Argon2 and PBKDF2.
so the one you used on every website, every ssh key, every gpg key, that 
is created with the same pattern you usually use on all password that 
leaked on the darknet? that i weak to dictonary attack? (just because 
you're human, and human construct itself on patterns it employ)
>
>> it's up to me to ensure it won´t... I've no such application that 
>> update exif/id3tag and also check for file corruption (or some 
>> intruder injection) via cron job... never detected any.
>
> I'm glad you haven't had this experience. As soon as you do and 
> discover your encrypted documents are now unreadable, you might change 
> your mind.
it's 15 year or more that I use this... yes it might end unreadable, but 
a ssd contrôler dying will have the same effet, a sector corruption on 
the header of a cipher storage will also...
>
>> if I cipher a file with a strong password created with information 
>> hashed from a file (that will likely create uniq data, since even a 
>> CRC32 has  1 in 4.29 billion chance to create the same crc... and 
>> crc32 is not comparable to sha256 and others) that can include even 
>> not type- able characters
>>
>> and then cipher the result with a weak password
>>
>> if an attacker try to bruteforce it, of course the weak password will 
>> be easily broken, but what the attacker get is a ciphered result. The 
>> attacker can't know if the password was guessed because the result is 
>> still a ciphered result.
>
> You've invented a primitive key derivation function.
>
> Why is this better than Argon2 or PBKDF2, are modern, peer-reviewed, 
> and successfully deployed key derivation functions?
i'm not, i'm just storing something in my brain that I can remember 
easely and feed to some "modern, peer-reviewed, and successfully 
deployed key derivation functions"
>
>> if you get arrested by the secret police, they will assume you hidden 
>> some sensitive information somewhere else in another ciphered file.
>
> Let's say, for sake of argument, you're a human rights worker in Gaza 
> who uses VeraCrypt to secure your laptop's hard drive. You get caught 
> up in an IDF dragnet. You fully cooperate with the IDF, showing them 
> all your data. There's nothing on it, nothing, to suggest you're a 
> terrorist. But the IDF still isn't going to let you go, because *how 
> can they know you're cooperating?*
>
> You can't prove to them that you're being honest. And that means 
> instead of being allowed to go free, you're going to spend time in 
> detention until some intelligence spook is able to make an assessment 
> of your risk level if released. It could be months.
>
> (No one should try to hijack this into an Israel-vs-Gaza discussion. 
> I'm sure many of us have strong feelings on the subject, but please 
> express them somewhere else.)
>
there are thousand of people sentenced to death without any single 
evidence, number of people sent to jail waiting for death execution and 
freed with a "mea culpa" with life impossible to recover (technically 
sending them to street life and death)... Sadely if some policemen 
decided you have to be the culpit, even if you fully cooperate, they 
will find what they will choose to be a proof, a corrupted file without 
header that you recovered, that contain random data, can be chosen to be 
a ciphered file.

how many people have been sentenced to death and executed during WWII 
without any evidence? just because you said hello to a suspicious guy 
that was suspected to be in the resistance?

sadely, even if you cooperate fully, you won't prove them that you're 
honest if they decided otherwise, just the fact you sent an email to 
this list, that you have some gpg key you played with... What you write 
today because you've some free speech possibilities can be used as an 
evidence elsewhere. I was questioned just because I went with some 
friends on a road trip in Morocco 9 years ago and that there was still a 
stamp on my passport, they asked be detailed information that I don't 
even remember, where exactly I stayed (it's a road trip, you decide 
where you stay day by day, with precise adress, phone, name of the 
persons I met...), and the more I couldn't say, the more they were 
digging and the only thing that saved me is the boarding of the plane. 
Stayed questionned 3 hours just because of some vacation I don't 
remember in detail but that was really nice and enjoyable. Did I prove 
to them I'm beeing honest? not sure, I just think that I won't go to the 
country that questionned me 3 hours without any reason anymore... 
technically I'm just suspicious if I just protect my data from beeing 
read by google and cipher it using gpg! Just because I don't like that 
someone that have no right to do it, eavesdrop, just because I store my 
vacation photo on a ciphered drive just in case it get stolled and I 
don't like the idea of having all my photos somewhere owned by someone I 
don't know. As long as I protected something, there is a non 
negligeable  possibility I stored some evidence somewhere else... be 
assured that if they decided, you will be the culpit. If they need a 
culpit, you'll be the culpit. If they want you to be a terrorist, you'll 
be a terrorist.

Even if you collaborate, the only way for someone to be sure that you 
didn't lied or hide some information is death (and even there, there is 
doubt) because you assume that at some point, a person will value more 
it's life than the information.
Even if nothing ciphered, all in clear text, there are lot of "justice 
mistake" with people sent to jail, without real evidence. Just because 
they were at the wrong time, wrong place, and that they have to be the 
culpit. How many people died just because of their origin?

so, ok, you can say veracrypt can put you to death, but this means that 
the threatener has enough knowledge to know there might be 2 way and 2 
information, which is not given, the only thing that can save you is 
buying enough time by giving them a decoy that seems enough valuable to 
take time to try it. If a person threaten you to gain access to your 
firm, he can still kill you after verifying he got the access, 
collaborating is not giving you the assurance he won´t do what you want 
to avoid.