On 9/24/26 00:20, Alexander Berg wrote:
Maarten is right on both counts, and I take them.
The defect detail belonged in the project's issue tracker, not on this
list.
The maintainer has the full material, and anything further I find goes
there.
The wording was the bigger mistake. I wrote in the tool's own internal
terms
without explaining a single one. If a paragraph only parses for the
author of
the code, it has no business on a list of several hundred people -
that is on
me, not on the reader.
One point from the original thread does belong here, in plain words.
The advice
we give users is "read the PKGBUILD", but what runs as root also
includes the
install scriptlet and any patch files committed next to the recipe. A
tool that
reads only the recipe matches the advice and misses the attack.
Whatever the
AUR ends up blessing, that is the gap worth closing.
Alexander Berg
This email makes me conflicted between laughing and getting angry. So
I'll just assume that this AI written apology is a joke and go with the
first option.
Seriously though, this kind of thing should not become normal. I really
don't need an apology from Claude. I actually don't need an apology at
all. My only point was that I don't want to talk to AI on the mailinglist.
In my opinion, the mailing list is for Arch people to talk with
each-other. For me, there is zero value (or sometimes even less) in
posting AI output verbatim as if it is your own message.
If you want to use AI as a tool to help talk in English, I think this
makes sense. then you I would prefer if you would write the message
yourself in your native language, and just let AI translate it. Then
at-least the message is truly yours.
Kind regards,
Maarten de Vries
